Post on 24-Jun-2020
Relationship marketing: an evaluation of trustworthiness within the Jordanian hotel sector Kharouf, H. Submitted version deposited in CURVE April 2011 Original citation: Kharouf, H. (2010) Relationship marketing: an evaluation of trustworthiness within the Jordanian hotel sector. Unpublished PhD Thesis. Coventry: Coventry University. Copyright © and Moral Rights are retained by the author. A copy can be downloaded for personal non-commercial research or study, without prior permission or charge. This item cannot be reproduced or quoted extensively from without first obtaining permission in writing from the copyright holder(s). The content must not be changed in any way or sold commercially in any format or medium without the formal permission of the copyright holders. Some materials have been removed from this thesis due to third party copyright, including appendices 3-9. The unabridged version of the thesis can be viewed at the Lanchester Library, Coventry University.
CURVE is the Institutional Repository for Coventry University http://curve.coventry.ac.uk/open
Relationship Marketing: An Evaluation of Trustworthiness Within the
Jordanian Hotel Sector
Husni Kharouf Doctoral of Philosophy (candidate)
Coventry University February 2010
A thesis is submitted in partial fulfilment of the University’s requirements for the Degree of Doctor of Philosophy. Coventry University - UK
ABSTRACT
The objective of this study is to investigate the concept of trustworthiness and
then examine its effect within the venue of the hotel sector. Given trustworthiness
accepted importance to relationship marketing, there appears to be a failure to
develop a coherent framework to indicate trustworthiness. This is a gap that this
thesis addresses and by doing so, it will extend the body of knowledge by contributing
to our understanding of the construct and its determinants.
The main hypothesis in this thesis is identifying the determinants of
trustworthiness with an outcome as part of a causal model. Within the proposed
model, the determinants; consistency, competence, integrity, benevolence, value
alignment and communication are assumed to have a positive impact on
trustworthiness. In turn, trustworthiness has a positive impact on both attitudinal
loyalty and behavioural loyalty, the two types of loyalty are proposed as the model
outcome, whereas the previous six determinants are proposed as antecedents of
trustworthiness.
In order to test the proposed hypotheses, a new measurement scale was
developed in order to evaluate trustworthiness with its determinants and outcome, the
model was tested within the hotel sector in Jordan where over 526 respondents took
part in the main survey collection, 60 respondents participated in the pilot study
along with 11 interviewees.
The results from the empirical study revealed that the hypothesised model is
valid and significant, in which all the antecedents of trustworthiness had a significant
loading as well as the model corollaries. These loadings vary in its significance and
strength; this created a clearer picture on the expected impact of each of these
determinants once the model being applied within service organisations.
ACKNOWLEDGMENT
I would like to acknowledge the contribution of the following people, without
whose help, guidance and motivation this thesis would not have been completed.
I would like to acknowledge the guidance and support from my two
supervisors, Professor Tom Donnelly and Dr. Harjit Sekhon. For their honest
guidance and continued support without which it would have been difficult if not
impossible to complete this thesis. I am indeed very proud and privileged to be
supervised by them.
Thanks to Tom and Harjit.
My heartfelt gratitude goes to my parents for ever believing in me.
THESIS RELATED LIST OF PUBLICATIONS
Kharouf, H., Sekhon, H., (TBA), Trustworthiness in Service Organisations: A
Structural Model, Journal of Service Industries, Under Review.
Kharouf, H., Scheer, L., (2009), Differences in the Nature and Role of Fairness in the
United Kingdom and Jordan, EMAC, Nantes, France.
Kharouf, H., Sekhon, H., (2008), An Evaluation of Trustworthiness within the Hotel
Sector: A Conceptual Framework and Scale Development, ANZMAC, Sydney.
Kharouf, H., Sekhon, H., (2008) Trustworthiness within the Hotel Sector: A
Structural Model, IIMA, Ahmad Abad, India.
Kharouf, H; Sekhon, H (2007), Trustworthiness in Hotel Services: A Conceptual
Framework, 1st Biannual International Conference “Strategic Developments in
Services Marketing”, Greece.
Kharouf, H; Sekhon, H, (2005), What Now That Services Can Walk Erect? SERSIG
Conference, Singapore.
Kharouf, H, (2004), Relationship marketing: A Cross Cultural Examination of Trust
in the Hotel Sector, Academy of Marketing Conference (doctoral colloquium),
UK.
Table of Contents
CHAPTER ONE: INTRODUCTION TO THE THESIS 1.1 Introduction ................................................................................................................... 1
1.2 Theoretical Background ................................................................................................ 1
1.3 Research Aim and Objectives ....................................................................................... 4
1.4 Research Hypotheses ..................................................................................................... 5
1.5 Structure of the Thesis .................................................................................................. 7
1.6 Research Approach ....................................................................................................... 9
1.7 The Nature of Service Organisations .......................................................................... 10
1.7.1 Relationship Marketing within a Services Context .................................................. 12
1.8 The Jordanian Hotel Sector ........................................................................................ 12
1.9 Contributions of the Study .......................................................................................... 15
Summary ........................................................................................................................... 17
CHAPTER TWO: THE DEVELOPMENT OF THE RELATIONSHIP MARKETING CONCEPT 2.1 Introduction ................................................................................................................. 18
2.2 Early Development of Marketing Thought ................................................................. 19
2.2.1 The Marketing Mix ................................................................................................. 21
2.3 The Emergence of Relationship Marketing ................................................................ 24
2.4 Theoretical Foundations of Relationship Marketing.................................................. 29
2.4.1 Business Marketing and The Interaction Network Approach ................................... 30
2.4.2 The Services Marketing Approach .......................................................................... 30
2.4.3 Marketing Channels ................................................................................................ 31
2.4.4 Database Marketing and Direct Marketing .............................................................. 32
2.5 Defining Relationship Marketing ................................................................................ 32
2.5.1 Relationship Marketing: A Paradigm Shift .............................................................. 37
2.5.2 Relationship Marketing as a European or American Model ..................................... 38
2.6 The Core Elements of Relationship Marketing .......................................................... 40
2.6.1 Trust ....................................................................................................................... 40
2.6.2 Commitment ........................................................................................................... 42
Summary ........................................................................................................................... 42
CHAPTER THREE: THE CONCEPT OF TRUST 3.1 Introduction ................................................................................................................. 43
3.2 The Concept of Trust .................................................................................................. 44
3.3 The Meaning of Trust .................................................................................................. 45
3.4 Different Perspectives on Trust ................................................................................. 51
3.4.1 Calculative (Economic) View of Trust .................................................................... 52
3.4.2 Social View of Trust ............................................................................................... 55
3.4.3 Psychological View of Trust ................................................................................... 56
3.4.4 Sociological View of Trust ..................................................................................... 57
3.5 Trust and Distrust ....................................................................................................... 61
3.6 Trustworthiness ........................................................................................................... 63
Summary ........................................................................................................................... 68
CHAPTER FOUR: JUSTIFICATION OF THE CONCEPTUAL MODEL AND THE RESEARCH HYPOTHESES 4.1 Introduction ................................................................................................................. 69
4.2 Characteristics of Trust............................................................................................... 70
4.3 Advantages of Trustworthiness ................................................................................... 75
4.4 Dimensions of Trust in The Literature ....................................................................... 77
4.5 The Production of Trust .............................................................................................. 81
4.6 Modelling Trustworthiness ......................................................................................... 90
4.6.1 The Proposed Model and Research Hypotheses ....................................................... 92
4.6.2 Trustworthiness ...................................................................................................... 93
4.6.3 Determinants of Trustworthiness Hypotheses .......................................................... 94
4.6.3.1 Low Level Determinants .................................................................................. 94
4.6.3.2 High Level Determinants ................................................................................. 95
4.6.4 Customer Loyalty (Model Outcome) ...................................................................... 98
Summary ......................................................................................................................... 100
CHAPTER FIVE: RESEARCH METHODOLOGY 5.1 Introduction ............................................................................................................... 101
5.2 The Research Philosophy .......................................................................................... 103
5.2.1 Choice of Philosophy ............................................................................................ 103
5.3 Research Logic .......................................................................................................... 108
5.3.1 Qualitative vs. Quantitative Methods..................................................................... 108
5.3.1.1 Quantitative Research .................................................................................... 109
5.3.1.2 Qualitative Research ...................................................................................... 110
5.3.1.3 Combined Methods/Triangulation .................................................................. 112
5.4 Data Collection .......................................................................................................... 114
5.4.1 Interviews ............................................................................................................. 115
5.4.2 Survey Questionnaires... ...................................................................................... 116
5.5 Sampling .................................................................................................................... 118
5.5.1 Sample Size .......................................................................................................... 119
5.6 Measurement Development ....................................................................................... 120
5.7 The C-OAR-SE Procedure for Scale Development .................................................. 122
5.7.7 Criticism of C-OAR-SE ........................................................................................ 128
5.8 The Structure of the Data Collection ...................................................................... 130
5.8.1 Internal Consistency ............................................................................................. 130
5.8.2 Unidimensionality Exploration.............................................................................. 131
5.8.3 Measure Validation ............................................................................................... 132
5.8.4 Normality Assessment .......................................................................................... 132
5.9 Quality of Research ................................................................................................... 133
5.9.1 Research Validity ................................................................................................. 133
5.9.1.1 Translation validity ........................................................................................ 134
5.9.1.2 Construct Validity Assessment ....................................................................... 134
5.9.1.3 Discriminant Validity Assessment .................................................................. 136
5.9.2 Reliability Assessment ......................................................................................... 137
5.10 Methods of Data Analysis ........................................................................................ 139
5.10.1 Structural Equation Modelling ............................................................................ 139
5.10.1.1 SEM vs. Regression ..................................................................................... 140
5.10.1.2 SEM: Assumption Criteria............................................................................ 140
5.10.1.3 Measurement Model ..................................................................................... 143
5.10.1.4 Structural Model........................................................................................... 144
5.10.2 Confirmatory Factor Analysis ............................................................................. 145
5.10.3 Statistical Methods for Hypotheses Testing ......................................................... 146
5.10.4 Estimation Method .............................................................................................. 148
5.10.4.1 Structured Models Comparison..................................................................... 148
5.10.5 Missing Data Treatment ...................................................................................... 149
5.10.6 SEM: Two-Step Method ..................................................................................... 149
5.11 Research Limitations ............................................................................................... 151
Summary ......................................................................................................................... 153
CHAPTER SIX: SCALE DEVELOPMENT AND PILOT STUTDY
6.1 Introduction ............................................................................................................... 154
6.2 Methods of Enquiry ................................................................................................... 155
6.3 Scale Development ..................................................................................................... 156
6.4 Overview of the Pilot Study ....................................................................................... 161
6.4.1 The Interviews ...................................................................................................... 163
6.4.2 The Questionnaire ................................................................................................. 166
6.5 Scale Evaluation ........................................................................................................ 169
Summary ......................................................................................................................... 174
CHAPTER SEVEN: DATA ANALYSIS AND RESULTS – DESCRIPTIVE ANALYSIS OF MEASUREMENT SCALE ITEMS
7.1 Introduction ............................................................................................................... 175
7.2 Data Collection Sample and Response Rate ............................................................. 176
7.2.1 Questionnaire Fatigue ........................................................................................... 177
7.2.2 Data Screening...................................................................................................... 177
7.2.3 Errors ................................................................................................................... 178
7.2.4 Missing Data ........................................................................................................ 179
7.2.5 Normality Tests .................................................................................................... 179
7.2.6 Outlier Identification ............................................................................................. 181
7.3 Descriptive Findings .................................................................................................. 182
7.3.1 Gender of Respondents ......................................................................................... 182
7.3.2 Nationality of the Sample ...................................................................................... 182
7.3.3 Age Groups .......................................................................................................... 183
7.3.4 Hotel Services....................................................................................................... 184
7.3.5 Purpose of the Visit............................................................................................... 184
7.3.6 Type of Used Service ............................................................................................ 185
7.3.7 Longevity of Customer Relationships.................................................................... 186
7.4 Means and Standard Deviation Analysis .................................................................. 187
7.4.1 Proposed Model Constructs................................................................................... 188
7.5 Additional Influential Factors ................................................................................... 196
7.6 Reliability and Validity of Measurement Scales ....................................................... 197
7.6.1 Reliability of Measurement Scales ........................................................................ 197
7.6.2 Validity of Measurement Scales ............................................................................ 199
Summary ......................................................................................................................... 200
CHAPTER EIGHT: MULTIVARIATE DATA ANALYSIS AND MODEL VALIDATION 8.1 Introduction ............................................................................................................... 201
8.1.1 Analysis Strategy .................................................................................................. 205
8.2 The Measurement Model .......................................................................................... 206
8.2.1 Testing the Measurement Model ........................................................................... 208
8.2.2 Overall Measurement Model ................................................................................. 211
8.3 Factor Analysis .......................................................................................................... 214
8.3.1 Exploratory Factor Analysis .................................................................................. 214
8.3.2 Confirmatory Factor Analysis ............................................................................... 218
8.3.3 Goodness of Fit Indices Discussion ....................................................................... 223
8.3.4 CFA with Calibration and Validation Sample ........................................................ 227
8.3.5 CFA analysis for the Model’s Constructs .............................................................. 228
8.3.6 Assessing Construct Validity ................................................................................ 237
8.4 The Structural Model Process ................................................................................... 239
8.4.1 Path Analysis ........................................................................................................ 241
8.4.2 Testing Alternative Models ................................................................................... 243
8.4.3 Evaluation of Fit for the Alternative Models ......................................................... 248
8.5 Hypotheses Testing .................................................................................................... 249
Summary ......................................................................................................................... 253
CHAPTER NINE: CONCLUSIONS AND MANAGERIAL IMPLICATIONS
9.1 Introduction ............................................................................................................... 254
9.2 General Findings and Discussion .............................................................................. 256
9.3 Research Hypotheses and Recommendations ........................................................... 259
9.4 Contribution of the Thesis ......................................................................................... 265
9.4.1 Theoretical and Methodological Contributions ...................................................... 266
9.5 Managerial Implications of the Research Findings .................................................. 269
9.6 Directions for Future Research ................................................................................. 271
9.7 Study Limitations ...................................................................................................... 272
REFERENCES ................................................................................................................ 275
APPENDICES ...................................................................................................................... i
List of Tables
Table 1.1: Number of hotels in Jordan ................................................................................. 14
Table 1.2: An overview of the Jordanian Hotel Sector ......................................................... 14
Table 2.1: Comparison between the Major Schools of Relationship Marketing .................... 26
Table 2.2: Key Relationship Marketing Definitions ............................................................. 34
Table 3.1: Key Definitions of Trust ..................................................................................... 47
Table 4.1: Classification of Trust by Discipline with Key Authors ....................................... 70
Table 4.2: Key Themes of Trust .......................................................................................... 80
Table 4.3: Mapping the Dimensions of Trustworthiness....................................................... 91
Table 5.1: Marketing Research: Main Scientific Paradigms and Their Elements ................ 104
Table 5.2: Assumptions of the Quantitative and Qualitative Research Paradigms ............... 109
Table 5.3: Differences between Qualitative and Quantitative Research Strategies .............. 111
Table 5.4: Different Research Methodologies .................................................................... 113
Table 5.5: Breakdown of the Research Approach............................................................... 121
Table 6.1: Main Empirical Studies on Trust and the Scales Used ....................................... 159
Table 6.2: Detailed view of the Data Analysis Strategy ...................................................... 162
Table 6.3: Examples of the Literature Supporting the Research Constructs ........................ 164
Table 6.4: The Final Research Items .................................................................................. 170
Table 7.1: Survey Response Rate ...................................................................................... 177
Table 7.2: 5% Trimmed Mean ........................................................................................... 180
Table 7.3: Tests of Normality ............................................................................................ 181
Table 7.4: Gender of Respondents ..................................................................................... 182
Table 7.5: Nationality of Respondents ............................................................................... 183
Table 7.6: Age Groups ...................................................................................................... 184
Table 7.7: Purpose of the Visit .......................................................................................... 185
Table 7.8: Type of Used Service ........................................................................................ 185
Table 7.9: Service Used vs. Purpose of Visit...................................................................... 186
Table 7.10: Length of the Relationship between Customers and the Hotel .......................... 187
Table 7.11: Frequency of Service Usage ............................................................................ 187
Table 7.12: Mean and Standard Deviation for the Model Constructs .................................. 188
Table 7.13: Mean and Standard Deviation for the Integrity Construct ................................ 189
Table 7.14: Mean and Standard Deviation for the Benevolence Construct .......................... 189
Table 7.15: Mean and Standard Deviation for the Competence Construct .......................... 190
Table 7.16: Mean and Standard Deviation for the Value Alignment Construct ................... 191
Table 7.17: Mean and Standard Deviation for the Consistency Construct ........................... 191
Table 7.18: Mean and Standard Deviation for the Communication Construct ..................... 192
Table 7.19: Mean and Standard Deviation for the Behavioural Loyalty Construct .............. 193
Table 7.20: Mean and Standard Deviation for the Attitudinal Loyalty Construct ................ 194
Table 7.21: Mean and Standard Deviation for the Trustworthiness Construct ..................... 195
Table 7.22: Brand Name vs. Gender .................................................................................. 196
Table 7.23: Security of the Hotel vs. Gender...................................................................... 196
Table 7.24: Reliability Analysis......................................................................................... 198
Table 8.1: Overview of the Analyses Strategy ................................................................... 205
Table 8.2: The Model Indicators (items) in Relation to the Model Constructs .................... 209
Table 8.3: KMO and Bartlett’s test of Sphericity ............................................................... 215
Table 8.4: Rotated Component Matrix ............................................................................... 217
Table 8.5: Factor Loading and Standard Error for the Constructs ....................................... 219
Table 8.6: Correlations between the Constructs ................................................................. 219
Table 8.7: Results of CFA for the Overall Measurement Model ......................................... 222
Table 8.8: Overall Model Fit Indices ................................................................................. 223
Table 8.9: Results of CFA for Calibration and Validation Split Samples ............................ 227
Table 8.10: Integrity Regression Weights .......................................................................... 229
Table 8.11: Competence Regression Weights .................................................................... 230
Table 8.12: Value alignment Regression Weights .............................................................. 231
Table 8.13: Communication Regression Weights ............................................................... 232
Table 8.14: Benevolence Regression Weights .................................................................... 233
Table 8.15: Consistency Regression Weights ..................................................................... 234
Table 8.16: Trustworthiness Regression Weights ............................................................... 235
Table 8.17: Behavioural Loyalty Regression Weights ........................................................ 236
Table 8.18: Attitudinal Loyalty Regression Weight ........................................................... 236
Table 8.19: Goodness of Fit Indices for the Alternative Models ......................................... 248
Table 8.20: Chi-square Comparison between the Models ................................................... 248
List of Figures Figure 1.1: The Research Focus of the Thesis in Relation to the Literature............................. 2
Figure 1.2: The Proposed Model and Hypotheses ................................................................. 6
Figure 2.1: The Development of Marketing Over the Past Sixty Years ................................ 20
Figure 2.2: The Disciplinary Roots of Relationship Marketing ............................................ 29
Figure 2.3: The Key Mediating Variables of Relationship Marketing ................................... 41
Figure 4.1: Antecedents and Outcomes of Trust ................................................................... 77
Figure 4.2: Mayer et al.’s (1995) Model of Trust ................................................................. 81
Figure 4.3: Various Relationship Forms............................................................................... 84
Figure 4.4 Forms of Dependence, Risk and Qualities of Trustworthiness ............................. 85
Figure 4.5: Trust Continuum ............................................................................................... 86
Figure 4.6: Classification of Different Trust Typologies ...................................................... 87
Figure 4.7: Trust and the Mediating Lens ............................................................................ 88
Figure 4.8: Conceptual Framework ...................................................................................... 93
Figure 5.1: Overview of the Methodology Chapter ............................................................ 102
Figure 5.2: Blocks of Research .......................................................................................... 103
Figure 5.3: The Key Research Paradigms .......................................................................... 106
Figure 5.4: Deductive and Inductive Logics ....................................................................... 108
Figure 5.5: Data Collection Process ................................................................................... 114
Figure 6.1: Overview of the Scale Development Process and the Pilot Study ..................... 155
Figure 8.1: The Process of SEM Analysis in This Chapter ................................................. 204
Figure 8.2: The Hypothesised Model and the Relationships between Constructs ................ 207
Figure 8.3: Part One of the Hypothesised Measurement Model .......................................... 210
Figure 8.4: Part Two of the Hypothesised Measurement Model ......................................... 211
Figure 8.5: Integrity Regression Weights ........................................................................... 229
Figure 8.6: Competence Regression Weights ..................................................................... 230
Figure 8.7: Value Alignment Regression Weights.............................................................. 231
Figure 8.8: Communication Regression Weights ............................................................... 232
Figure 8.9: Benevolence Regression Weights .................................................................... 233
Figure 8.10: Consistency Regression Weights ................................................................... 234
Figure 8.11: Trustworthiness, Attitudinal and Behavioural Loyalty Regression Weight...... 235
Figure 8.12: Complete Path Diagram with the Hypothesised Measurement Specifications . 240
Figure 8.13: The Path Loadings Diagram........................................................................... 242
Figure 8.14: Alternative model A ...................................................................................... 244
Figure 8.14: Alternative model B ...................................................................................... 245
Figure 8.14: Alternative model C ...................................................................................... 246
Figure 8.14: Alternative model D ...................................................................................... 247
Chapter One – Introduction
-1-
CHAPTER ONE: INTRODUCTION TO THE
THESIS
1.1 Introduction
Service organisations invest vast sums of money in the development of their
customer relationships in order to improve their overall business performance.
Nevertheless, there is limited academic and practitioner research regarding how these
‘customer–organisation’ relationships’ can be built and maintained, explicitly in terms
of relationship trustworthiness. This thesis aims to explore relationship
trustworthiness and its determinants within the arena of the Jordanian hotel sector.
Firstly, this chapter will present the aims and objectives of the research.
Secondly, it will develop the rationale of the research initiative, and finally, it will
outline the structure of the thesis. Section 1.2 provides a brief theoretical background
for the research in the thesis. Section 1.3 explains the specific aim and objectives of
the thesis, while section 1.5 outlines the structure of the thesis and provides a
summary of the chapters to follow with their intended objectives and main research
contributions. An overview of the approach to the research is presented in section 1.6,
followed by the nature of service organisations in section 1.7, and then the overall
contribution to knowledge gained from the research is presented in section 1.8.
1.2 Theoretical Background
Emerging from both the empirical and theoretical debates surrounding
relationship marketing, the concept of trust is accepted as a key component in
successful relationships (Mayer et al. 1995; Doney and Cannon 1997), and it is often
seen as an important factor in defining the strength of customer relationships (Morgan
and Hunt 1994). However, most of the current literature focuses on trust as an
antecedent of customer loyalty (Buttle and Burton 2002; Bartelt 2002), effective
communication (Anderson and Narus 1990), customer satisfaction (Jyh-Shen 2004),
Chapter One – Introduction
-2-
customer commitment (Morgan and Hunt 1994) and risk reduction (Mayer et al.
1995).
Trust can be defined1
Source: Author
as “confidence in an exchange partner’s reliability and
integrity” (Morgan and Hunt 1994:23). Similarly, it is also understood in terms of
“the perceived credibility and benevolence of a target of trust” according to Doney
and Cannon (1997:36). The importance of trust emerges from its presence as a central
factor for all organisational transactions (Dasgupta 1988), reducing costs (Williams
1993), influencing the coordination of institutional organisations (Shapiro 1987) and
enhancing the organisational decision making process (McAllister 1995). Figure 1.1
provides an illustration of how the concept of trust integrates with both relationship
marketing and services marketing literature, a model that forms the theoretical basis
for this thesis.
1 A full discussion of the various definitions of trust in the literature can be found in Chapter Three - Section 3.3.
Relationship Marketing
Services Marketing
Figure 1.1: The Research Focus of the Thesis in Relation to the Literature
Relationship
Trust
Chapter One – Introduction
-3-
In spite of the fact that trust in service organisations attracts considerable
interest within various streams of research and related literature (see for example,
Butler and Cantrell 1984; Lewicki et al. 1998), most authors draw upon a belief-based
conceptualisation of trust; that is, perceived trustworthiness (Gefen et al. 2003). By
this definition, the deciding factor in the creation of trust is trustworthiness, which is a
characteristic of the service provider (the trustee). Therefore, trustworthiness is a
subjectively determined value judgment about the behaviour of the party that is
seeking to be trusted (Glaeser et al. 1999; Bews and Rossouw 2002; Caldwell and
Clapham 2003).
Trustworthiness2
Most literature sources use the term trust, although sometimes the focus is on
customer behaviour and sometimes on service provider attributes; the two foci are
even used interchangeably. This is well described by Hardin (2002:29) when he notes
that “much of the literature on trust hardly mentions trustworthiness, even though
implicitly much of it is primarily about trustworthiness, not about trust”. Hardin’s
(2002) argument was further confirmed by Caldwell and Clapham (2003) when they
argued that the trust decision is evaluated upon one party’s assessment of the
behaviours of another party, through what they described as a ‘mediating lens’ or
complex filter through which each person views the world. In other words, in order
for one party to trust, the trustee has to possess certain characteristics perceived as
being trustworthy.
is described in the literature as the perceived probability that
a particular trustee will maintain one’s trust (McKnight et al. 2002). Hardin (2002:28)
defines trustworthiness by stating that “… your trustworthiness is your commitment
to fulfil another’s trust in you”. Putting Hardin’s (2002) statement into context,
trusting beliefs represent a “… sentiment or expectation about an exchange partner’s
trustworthiness” (Moorman et al. 1993:315). Mayer et al. (1995) defined
trustworthiness as an attribute of a trustee that is responsible for trust. According to
Mayer et al. (1995), the perceptions of the trustor that affect their judgements about
the trustee concern the trustee’s ability, integrity and benevolence. Doney and Cannon
(1997) treat trust as a single construct incorporating trustworthiness, integrity, ability,
and benevolence.
2 Full details on the discussion surrounding trustworthiness can be found in Chapter Three-Section 3.6.
Chapter One – Introduction
-4-
Trustworthiness, however, is a more salient issue with high-risk transactions
(for example when purchasing an accommodation in a different country). It is
important to conceptualise and distinguish trustworthiness from trust because: 1) trust
may be an enduring condition of the customer, i.e. relying on the customer to initiate
trust or distrust, with the implication that he/she is ultimately making the trust
decision, 2) trust is also an outcome of the evaluation of the trustworthiness of the
service provider. Hence, by refocusing on the strategic organisational value of
developing a trustworthy attribute(s), it is easier for the service provider to be able to
influence and predict trust, particularly in a new business context.
1.3 Research Aim and Objectives
While the issue of trust attracts significant attention within the literature in a
marketing sense, the theoretical understanding of the construct is both varied and
ambiguous. In addition to this ambiguity, a gap exists in terms of our understanding of
the notion of trustworthiness. This lack of understanding was recognised by Hardin
(2002:29) when he noted that “… much of the literature on trust hardly mentions
trustworthiness, even though implicitly much of it is primarily about trustworthiness,
not about trust”.
Given trust’s accepted importance within relationship marketing (see for
instance, Morgan and Hunt 1994), there appears to be a significant research gap
regarding a coherent framework for trustworthiness in the literature. This is a subject
addressed by this thesis; by doing so, this research will extend the body of knowledge
by contributing to our understanding of the construct of trustworthiness and its
attributes (for example, competence and value alignment).
Chapter One – Introduction
-5-
This thesis aims: to explore relationship trustworthiness and its determinants
within the arena of the Jordanian hotel sector.
The primary objectives of the thesis are:
1. To develop a theoretically grounded framework and develop a scale to
measure the determinants of trustworthiness and its outcome.
2. To investigate the factors which influence the determination of trustworthiness
between a service provider and its customers.
3. To carry out an empirical examination of the proposed model in the Jordanian
hotel sector.
While there are numerous studies that have examined trustworthiness within
the service sector, the main purpose of this thesis is to conceptualise and test the
determinants of trustworthiness. The results of this thesis will provide insights into the
importance of trustworthiness in the service sector, and will, therefore, provide a solid
base from which to establish better customer relationships within service
organisations. This, in turn, will enhance the overall service stability in the
organisation’s operations, and ultimately increase customer loyalty.
1.4 Research Hypotheses
Based on the research aim and objectives stated above, nine hypotheses are
proposed and a structural model will be developed and tested. The research in this
thesis takes the stance that modelling trust and rationalising the proposed framework
can be seen as establishing the foundation of trustworthiness by identifying the
antecedents of trust – in other words, those aspects of trust that ultimately lead to
trustworthiness. Customer loyalty is conceptualised in the model as an outcome of
trustworthiness.
Initially, the proposed model (see Figure 1.2) assumes three main relationships
and clearly defines the determinants of trustworthiness as well as the outcome of
customer loyalty. The main relationships in the model place trustworthiness at the
focal point, with various determinants of trustworthiness (classified as low and high
level) having a direct relationship with the central characteristic. The importance of
Chapter One – Introduction
-6-
each determinant will be discussed at a later stage in this thesis, since the results from
the empirical study will show which determinant is the most significant, and having
the strongest regression coefficient paths with trustworthiness. The third part of the
model involves the outcome of trustworthiness, measured by two types of customer
loyalty; behavioural (low level) and attitudinal (high level) loyalty. The model also
includes a prediction that behavioural loyalty can lead to attitudinal loyalty.
As drawn from the above diagram (Figure 1.2), the research hypotheses are:
H1: Consistency has a positive impact on trustworthiness.
H2: Competence has a positive impact on trustworthiness.
H3: Integrity has a positive impact on trustworthiness.
H4: Benevolence has a positive impact on trustworthiness.
H5: Value Alignment has a positive impact on trustworthiness.
Benevolence
Competence
Integrity
Consistency
Value alignment
Communication
Behavioural loyalty
Attitudinal loyalty
Trustworthiness
Low level
High level
+
+
+
+ +
+
+
+ +
Figure 1.2: The Proposed Model and Hypotheses
Source: Author
Chapter One – Introduction
-7-
H6: Communication has a positive impact on trustworthiness.
H7: Trustworthiness has a positive impact on attitudinal loyalty.
H8: Trustworthiness has a positive impact on behavioural loyalty.
H9: Behavioural loyalty has a positive impact on attitudinal loyalty (a prediction).
These hypotheses will be further discussed and justified based on the relevant
literature in Chapter Four section 4.6.1, and will be tested empirically in Chapter
Eight.
1.5 Structure of the Thesis
To begin with, the thesis provides a brief introduction detailing the research
setting and a theoretical background for the study (Chapter One). This introductory
chapter is followed by a review of the literature surrounding relationship marketing
(Chapter Two) and the concepts of trust and trustworthiness along with their various
dimensions (Chapter Three). The major benefits of trust for a service organisation will
be underlined, followed by a discussion of how trustworthiness is conceptualised
(Chapter Four). Following this chapter, a detailed research methodology will be
discussed (Chapter Five). The fieldwork and discussion of the research findings are
presented in three chapters: the scale development and the pilot study (Chapter Six);
the descriptive findings (Chapter Seven); and a multivariate statistical analysis of the
proposed model (Chapter Eight). The final chapter outlines the main conclusions
based on the research findings. In addition, this chapter includes a discussion of the
managerial implications of the research, along with its limitations of the research
(Chapter Nine). In more detail, the structure of the thesis is as follows:
Chapter One – Introduction to the Thesis
This chapter presents the aims and objectives of the thesis, demonstrating the
significance of the study and presenting the structure of the thesis. The chapter
provides an introduction to the research, as well as a brief overview of the approach
and its main parameters.
Chapter One – Introduction
-8-
Chapter Two – The Development of the Relationship Marketing Concept
This chapter introduces the concept of relationship marketing and
demonstrates how relationship marketing has become a crucial part of the marketing
domain. In addition, the chapter identifies how the concept of trust emerged in the
literature and how it integrates with various marketing concepts (for example, the
logic of service dominance).
Chapter Three – The Concept of Trust
This chapter provides a full discussion on the concept of trustworthiness and
shows its development within the literature, including a variety of definitions,
concepts and theoretical models within different academic schools of thought.
Furthermore, a necessary distinction between trust and trustworthiness is discussed.
Chapter Four – Justifications of the Conceptual Model and the Research Hypotheses
The conceptual model and the research hypotheses are discussed during this
chapter. In addition, the chapter identifies the antecedents of trustworthiness along
with their effects as part of the proposed theoretical model. Nine hypotheses are
developed and discussed within the model.
Chapter Five – Research Methodology
This chapter aims to provide an overview of the research design and approach,
and to outline the data collection strategy for the empirical part of the study. It also
justifies the data analysis strategy. A detailed explanation of the phases of the research
(a qualitative study, expert judging, a pilot and finally the main study) is presented,
along with a justification of the analysis stages (new measure development and
validation, followed by assessment of the structural equation model). The discussion
includes the reasons for using structural equation modelling as a key analysis tool to
test the proposed model.
Chapter One – Introduction
-9-
Chapter Six – Scale Development and Pilot Study
This chapter provides an overview of the scale development process and a
discussion of how the main research items have been generated and used to construct
the research instrument and test the proposed model. The chapter also reports the
results of the pilot phase of the research, and discusses the reliability of the proposed
scale.
Chapter Seven – Descriptive Analysis of Measurement Scale Items
The aim of this chapter is to provide early results from the collected data and
to examine the nature of the selected sample. There is a detailed discussion of how the
raw data are treated and purified, and several descriptive analytical tests on the sample
are also discussed.
Chapter Eight – Multivariate Data Analysis and Model Validation
All the analytical testing steps applied to the data, including the statistical
validation and the structural path analysis, are evaluated in this chapter. In addition,
the key results and a review of each hypothesis are fully examined.
Chapter Nine – Main Conclusions and Managerial Implications
This chapter discusses various considerations about the theoretical and
managerial implications of the research results. Furthermore, the main contributions
of the research are identified and discussed. The chapter also includes a reflection on
the limitations of the study, and a discussion of future research.
1.6 Research Approach
After defining the concept of trustworthiness and clarifying its impact within
service organisations, a critical realist approach was adopted. In the light of a multi-
disciplinary review of the literature and after an initial qualitative study, nine
hypotheses were developed and tested using a quantitative design.
Following the use of three expert judges (during the measure development
phase) and a pilot study (60 respondents), a further 529 questionnaires were collected
Chapter One – Introduction
-10-
for the main empirical study. They answered a self-administered questionnaire
designed to measure the proposed model. Subsequently, a structural equation
modelling analysis technique was used to validate the measures and test the
determinants and outcomes of trustworthiness.
1.7 The Nature of Service Organisations
Service organisations traditionally emerged from the industrial sector,
supporting the production process by distributing products and ensuring sufficient
profit at the same time (Zeithaml et al. 1985). However, the growth of service
organisations influenced the range of the available services in different sectors, for
example, in the hospitality sector, customers are clearly the main focus of the
business, and, therefore, delivering the right service becomes the most important
criterion for successful service organisations. within service organisations a need for
services marketing with an internal focus on service delivery was developed, as a
method to increase customer base and compete with market rivals; this has been
widely discussed in the marketing literature (see for example, Gummesson 1994;
Piercy and Morgan 1989, 1991; Piercy 1995).
The notion of services marketing gathered momentum during the 1970s when
several authors described its development (see for example, Grönroos 1978 and 1994;
Fisk et al. 1994)3
3 Appendix Two provides a full elaboration on the milestones in the services marketing literature.
. The 1980s witnessed the development of the notion of service
quality (Zeithaml et al. 1985), since the perceived quality of a service will be the
result of an appraisal procedure, comparing the perception of service delivery and its
results against customers’ expectations via a technical and functional quality outcome
process (Grönroos 1993). However, academics, practitioners and customers all attach
different meanings to the concept of services (Johns 1999); in an attempt to clarify
this. Several new definitions for services marketing appeared in the marketing
literature; for example, Mathe and Shapiro (1993:33) define services marketing as:
“Service is all of the activities undertaken by the firm to provide value in use over time, measured by increased customer satisfaction with a tangible product or series of products.”
Chapter One – Introduction
-11-
However, Mathe and Shapiro’s (1993) definition does not provide a clear
indication of the characteristics of services. Furthermore, a simple increase in
customer satisfaction is not a clear measure of service quality because other factors
play a role in affecting customer satisfaction (for example, ease and efficacy of
communication). Therefore, this thesis employs Johns’ (1999:961) definition of
services marketing because it takes into account several other intangible
characteristics of the service provider:
“Service as a process not only is the delivery of a core service, but also has a style or manner of its own imbued with artistic, dramatic or craftsman-like possibilities.”
Service encounters further influenced the integration of relationship marketing
within services marketing. Bitner (1990) argued from a relationship perspective that
the individual service encounter is seen as only one element in an ongoing sequence
of relationship episodes that together form a process between the customer and the
service provider. This concept emerged when marketing academics discussed the
dyadic contact between a service provider and customers, and it is identified as a
significant determinant of the customer’s general satisfaction with a service (Solomon
et al. 1985). Bitner (1990) indicated that a service encounter can be seen as a period
of time during which a consumer directly interacts with a service.
As has been proposed by Fisk et al. (1994), there are three main factors that
inform the service encounter theory, namely: internal marketing; service design and
relationship marketing. The outcome(s) of these concepts are often perceived as
service quality with a direct influence on the service encounter. From this argument it
can be concluded that relationship marketing plays a key role within services
marketing, and, hence, it is acceptable to discuss relationship marketing within service
organisations. This claim will be discussed below.
Chapter One – Introduction
-12-
1.7.1 Relationship Marketing within a Service Context
By the end of the 1980s services marketing had been formally accepted as a
new paradigm within the field of marketing, an area with different unique
characteristics from other streams of marketing research (for example, industrial
marketing; see Grönroos 1998; Zeithaml and Bitner 1996). Before long, academics
started to apply services marketing research to the industrial (goods) sector in an
attempt to study buyer-seller relationships within industrial marketing. This
represented a significant shift within the general services marketing paradigm, and
developed into what we know today as relationship marketing4
As mentioned earlier the research venue of this thesis is the Jordanian hotel
sector. In recent decades the Jordanian tourism sector has played an important role in
the Kingdom’s economy. The Jordanian hotel sector has grown by about seven
percent a year since the early 1980s, comprising on average around eleven per cent of
(Gummesson 1991;
Grönroos 2000; Christopher et al. 1991). Hence, relationship marketing emerged from
the conjunction between service marketing, industrial marketing and data base/direct
marketing (Möller and Halinen 2000).
Still, it can be argued that services marketing has benefited most from the
development of relationship marketing, since it has been suggested that relationship
marketing is best viewed in the context of service organisations. This is evident in
early work conducted by Crosby and Stephens (1987), who viewed relationship
marketing as an instrument that adds value to the product or service by meeting
certain peripheral demands (for example, making customers feel welcomed).
However, these authors argued that customers are mainly concerned with the core
product and service quality. Zineldin (1999) supported this stance when he argued
that the need for relationship marketing emerged because most service organisations
offer almost the same core service(s); therefore, differentiation will be of the greatest
interest to organisations with the strongest capability to develop long-term customer
relationships.
1.8 The Jordanian Hotel Sector
4 A full discussion of the emergence of relationship marketing can be found in Chapter Two, Section 2.3.
Chapter One – Introduction
-13-
the gross domestic economy. The Jordanian tourist industry exploits the wide range of
tourist attractions and activities that are situated in Jordan, such as the Dead Sea,
Wadi Rum, religious sites at Mount Nebo and al-Mazar, and mosques in Amman and
Madaba. Petra’s spectacular Nabatean ruins, Roman theatres, tombs and monasteries
are a major attraction for tourists (Jordanian Ministry of Tourism and Antiquities
2008).
Following a surge in tourism related to the 1994 peace agreement with Israel,
Jordan’s hotel occupancy rate soared unexpectedly. Major hotel expansion plans have
moved ahead in the hope that regional tensions will subside and lead to an increase of
non-Arab tourists. However, as around fifty percent or more of Jordanian tourist
arrivals are from within the immediate region, they are less disposed to have poor
perceptions in regard to the regional crises (for instance, political problems in the
neighbouring countries). Regardless of the uncertainty caused by regional instability
the number of visitors, based on diversified sources, should follow long-term growth
patterns. The overall trend is clearly positive, with hotel room occupancy up sixty
percent during the period 2000–2008, and the number of visitors increasing from 3.9
million in 2003 to 5.1 million in 2006. Overall, the national tourism strategy expects
the sector to contribute more than £1 billion annually to the economy, by 2010, and to
create fifty thousand jobs (Jordanian Ministry of Tourism and Antiquities 2008).
Over the past eight years Jordan has entered a period of comprehensive
political, social, and economic reform with the aim of building a modern state based
on economic vitality with substantial potential for growth and prosperity, political
inclusion and social stability. Jordan has made great strides in opening up and
liberalising its economy in order to accomplish sustainable economic development,
prompting export-led growth, and enabling the creation of a favourable environment
for inwards FDI. This can be seen clearly in the hotel sector, in which one hundred
percent of the five star hotels are owned by international chains and eighty percent of
the four star hotels are either partially or fully internationally owned. Despite
continuing regional instability, this sustained international investment is evident from
the growth in the Jordanian economy, which remains strong. The fastest growing
sectors in 2007 were: manufacturing industry (with an increase of 16.7%); hotels
Chapter One – Introduction
-14-
(13.1%); telecommunications and transport (11.8%); construction (11.7%); electricity
and water (9.9%); and the financial and real estate sectors (9.4%).
Table 1.1: Number of Hotels in Jordan
(Source: Jordanian Hotel and Tourism Association 2008)
The Jordanian hotel sector currently contains 474 hotels ranging in quality
from five star to one star accommodation (see Table 1.1)5. These hotels are spread
across the country, but 81% of the hotels are located in the capital city Amman (see
Table 1.2). Examining the hotel classification reveals that five star hotels dominate
the sector in terms of nights spent (see Table 1.2), which indicates the importance of
this category of accommodation and suggests the benefit of further improving this
specific category. Therefore, this study concentrates on this sector of the Jordanian
tourism industry, focusing on five star hotels6
Table 1.2: An Overview of the Jordanian Hotel Sector
.
(Source: Jordanian Hotel and Tourism Association 2008)
5 These numbers include hotel apartments, hostels and unclassified hotels, which this study did not cover. Source: Jordanian Ministry of Tourism and Antiquities 2008. 6 The majority of the five star hotels in Jordan (90%) are franchised based. Source: Jordanian Ministry of Tourism and Antiquities 2008.
Chapter One – Introduction
-15-
With average tourist numbers around 6.5 million in 2007 (including day
trippers and overnight visitors), and with projections for this figure to climb by 12%
in 2009 and 2010, the scene is set for business in the hotel sector to soar, backed up
by complementary marketing by the Jordan Tourism Board (JTB) and private sector
bodies such as the Jordan Incoming Tour Operators’ Association (JITOA). Enhancing
visitor relationships is vital for the sector’s growth. Such enhancement will not only
improve the level of customer satisfaction, but it will also contribute to building long-
lasting relationships by creating high levels of trustworthiness for the service
provider. This thesis applies the proposed model for trustworthiness within the
Jordanian hotel sector, aiming to contribute to both service organisations in general
and hotels in particular.
1.9 Contribution of the Study
The potential contributions of this study can be discussed from both a
theoretical and a practical standpoint. Essentially, this thesis contributes to an increase
in the current level of theoretical knowledge in the literature on trustworthiness. This
is an important field of study because trustworthiness is a key element in enhancing
and maintaining long-term relationships between the organisation and its customers
(Spekman 1988), to the benefit of both involved parties as trustworthiness evolves
(Hardin 2002)7
In detail, the particular theoretical contribution of the thesis is wide-ranging.
The established literature within the field of relationship marketing, for instance,
research by Morgan and Hunt (1994) and Doney and Cannon (1997) focuses on the
consequences of trust rather than its antecedents. Despite a more recent study by
Sirdeshmukh et al. (2002) dealing with the notion of trustworthiness, the literature
generally fails to make a distinction between the various types of trustworthiness, i.e.
higher and lower forms. The higher form of trustworthiness relates to notions of
benevolent behaviour and shared values, whereas the lower level is associated with
. This contribution is achieved by exploring the various literatures
surrounding the construct, and the development of a model to conceptualise the
concept within service organisations and test it in within the Jordanian hotel sector.
7 A detailed discussion on trustworthiness is presented in Chapter Three – Section 3.6.
Chapter One – Introduction
-16-
traits such as competence, effective communication, integrity, consistency and so
forth.
This thesis argues that higher level trustworthiness, allows the service provider
to add more value to the provided service than do lower levels. The proposed model
suggests that trustworthiness has different determinants; this addresses the existing
gap in the literature and by doing so, it provides service organisations with the
necessary information on how to apply and enhance their trustworthiness with their
customers, which will result in reducing operational costs, widening the customer
base and enhancing the organisation decision making process. Furthermore, the thesis
provides a methodological contribution by developing and empirically testing a new
scale to measure trustworthiness. The scale development comprises generated and
validating a pool of items to measure the research constructs8
.
Although it may be possible to debate whether trustworthiness is an attribute
or construct, for the purposes of this thesis it is treated as a construct, which
Diamantopoulos (2005) defines as an abstract entity. The features that allow an
organisation to be perceived as trustworthy, i.e. the determinants of trustworthiness
are the attributes that are conceptualised as part of the theoretical model.
In terms of a practical contribution, the findings of the study should provide
several insights about building trustworthiness in relationships, as well as customer
loyalty. The analysis and findings address the issue of clarity within the
conceptualisation of relationship trust in the literature, and the meta-analysis of the
theorised relationships includes a discussion of the construct’s impact and
significance.
Finally, the systematic examination of structural relationships in the attributes
helps to facilitate a clearer understanding of the nature of trustworthiness, its
antecedents and outcomes. The results could provide information about building
trustworthiness within service organisations, which might aid organisations in
enhancing their customer relationship practices.
8 See Chapter Six – Section 6.3 for a full discussion on the scale development stages.
Chapter One – Introduction
-17-
Summary
This chapter has presented the rationale underlying the research initiative that
forms the subject of this thesis. The main research aim (to explore the concept of
trustworthiness and its determinants within the venue of the Jordanian hotel sector)
has been examined, along with the three main objectives around which the research
and the thesis are structured. An outline of the thesis structure has been presented,
showing how each chapter contributes to the main objectives. Finally, the chapter has
presented an overview of the research approach and its main parameters.
Chapter Two – The Development of the Relationship Marketing Concept
-18-
CHAPTER TWO: THE DEVELOPMENT OF
THE RELATIONSHIP MARKETING
CONCEPT
2.1 Introduction
Several theorists contributed to the early development of marketing thought.
Harvey et al. (1996) summarised these contributions as: the marketing concept, the
marketing mix, and the generic marketing concept. Earlier, Webster (1992) argued
that it was not until the 1950s that an increased emphasis on marketing management
first appeared in the literature. Writing in 1972, Kotler was among the first to identify
changes in the academic categorisation of marketing. He suggested that marketing had
started originally as a branch of applied economics, before being embraced as a
management discipline with the objective of engineering increased sales. Kotler
(1972) also suggested, at that time, that marketing could be classified as an applied
behavioural science, since there was no emphasis on understanding buyer-seller
systems within the literature (Kotler 1972). Marketing practice has reacted to
changing economic environments, and today marketing is still an evolving subject
with a great deal of academic input1
The notion of relationship marketing has been debated several times in the
literature focusing upon the shortfalls in the traditional concept of the marketing mix
(Grönroos 1994), and has gone from being a novel paradigm shift to a new term for
an old phenomenon
.
2
1 See appendix Three for the key factors contributing to the marketing concept. 2 This is also a phrase Petrof (1998) used in his work to describe the development of relationship marketing.
(Petrof 1998). Moreover, marketing academics have debated the
extent to which relationship marketing forms a unique concept within the marketing
literature (for example, see Grönroos 1994). They also continue to question the
relevance of such a concept in replacing the traditional marketing mix, while at the
same time providing another dimension to manage relationships with customers.
Chapter Two – The Development of the Relationship Marketing Concept
-19-
Regardless of the extended debates in the literature, few would disagree about the
benefits of practising relationship marketing within organisations (for example,
improving customers’ loyalty).
This chapter explores the early development of marketing thought; more
specifically, it then discusses the different concepts and marketing approaches that
contributed to the emergence of relationship marketing. It will further explain how the
focus of the marketing literature has shifted towards the emergence of relationship
marketing as a method to establish and maintain a long-term customer relationship to
create a competitive advantage3
2.2 Early Development of Marketing Thought
for the organisation.
4
Webster (1988; 1992) argued that marketing gradually evolved in the early
1950s to become a management discipline devoted to increasing a company’s sales.
At that time marketing academics
5
Between the 1950s and 1980s (see Figure 2.1), the focus of marketing shifted
from basic goods and services to a broader perspective with more emphasis on
decision-making and problem solving, placing customers as the central focus for all
organisational activities (Vargo and Lusch 2004). Four main approaches were linked
to marketing at the time, namely: the commodity approach (manufactured goods and
services); the institutional approach (wholesaler agents and retailers); the managerial
approach (analysis, planning and organising); and the social or environmental
approach (marketing effectiveness, quality and social impact) (Kotler 1972). Each of
these approaches expanded and formed a separate field within marketing and
characterised marketing as an applied behavioural
science that consisted of understanding buyer and seller behaviours in the marketing
of goods and services. According to Vargo and Lusch (2004), the logic of marketing,
at that time, was a focus on transactions and output, and how organisations performed
marketing functions to add value to commodities (see also Levitt 1960; Webster 1992;
Vargo and Lusch 2004).
3 For a detailed discussion of competitive advantage please see Porter (1985); the research in this thesis adopts his definition. 4 See appendix Five for a full review on the consumer marketing theory. 5 For a full discussion on the development of marketing see Vargo and Lusch (2004).
Chapter Two – The Development of the Relationship Marketing Concept
-20-
contributed to the overall development of marketing theory (Kotler 1972). Figure 2.1
illustrates the development of marketing over the past sixty years.
Figure 2.1: The Development of Marketing over the Past Sixty Years
1950s 1960s 1970s 1980s 1990s 2000s
Consumer Marketing
(mass marketing)
Business to business
marketing
Non-profit and
societal marketing
Services marketing
Relationship marketing
Marketing as a Social and Economic Process –
relationship marketing
(Source: Adapted from Christopher et al. 1991; Vargo and Lusch 2004)
The development of marketing is clearly expressed in the definitions published
by various marketing institutions. For instance, the Chartered Institute of Marketing
(CIM) defined marketing in 2005 as follows:
“Marketing is the management process responsible for identifying, anticipating and satisfying customer requirements profitably.”
Another definition of marketing was offered by the American Marketing
Association (AMA), viewing marketing as a social process along the lines suggested
by Vargo and Lusch (2004) (see Figure 2.1). This is clearly reflected within the 2007
AMA definition:
“Marketing is the activity, set of institutions and process for creating, communicating, delivering, and exchanging offerings that have value for customers, clients, partners, and society at large.”
It can be noted from these two often-quoted definitions of marketing that the
American and the UK schools of thought are in contrast. In the UK, the CIM views
marketing as a management process responsible for meeting customers’ requirements,
while in the US the AMA considers marketing to be an activity involving set of
institutions, aiming to offer value to customers, partners, clients and society. This
thesis adopts the CIM definition as a working definition because of its focus on the
Chapter Two – The Development of the Relationship Marketing Concept
-21-
management process of marketing, and its emphasis on satisfying customers, an
indication of the importance of building long-term relationships.
The direction of marketing in terms of the shift in its focus and purposes was
criticised by a number of academics (for instance, see Vargo and Lusch 2004).
Earlier, Levitt (1960) stated that the marketing concept at that time was ‘myopic’
because it was basically product oriented instead of customer oriented, and it did not
consider the market environment. Therefore, an alternative model arose in the
marketing literature to overcome the ‘myopic’ concept of marketing. This new
concept became known as the marketing mix.
2.2.1 The Marketing Mix
The development of the ‘marketing mix’ came about in the early 1950s6 via
Borden’s (1964) notion of ‘marketing ingredients’ or marketing as being a ‘mixer of
ingredients’. Borden listed twelve such ingredients, namely: product planning,
pricing, branding, distribution channels, personal selling, advertising, promotions,
packaging, display, servicing, physical handling, and fact finding and analysis7
The marketing mix concept was reintroduced by McCarthy in the early 1960s.
This reintroduction was based on managing the four Ps (Product, Price, Place and
Promotion). McCarthy’s (1964) work summarised the
.
Borden tried to combine the twelve ingredients, based on a suggestion from Culliton
(1948), into a ‘marketing mix’ that would increase an organisation’s profitability
(Grönroos 1994).
8
6 Borden discussed this mixture of ingredients in his article ‘The Concept of the Marketing Mix’ in 1964. 7 For a more detailed discussion on the original twelve ingredients of the marketing mix, see Grönroos (1994). 8 Even though Borden published his article on the marketing mix in 1964, he addressed the concept of the marketing mix on several occasions, for example in 1953 in his American Marketing Association presidential speech.
notion of ‘marketing
ingredients’. Soon, this concept became the most popular classification continuum for
the marketing mix, because it covered all aspects of marketing at the time. Kotler
(1972; 2003) regards the mix principally as an optimisation based theory, providing
suggestions about how to achieve the most advantageous mix of the four Ps that act as
controllable parameters, likely to influence consumer buying and decision-making
Chapter Two – The Development of the Relationship Marketing Concept
-22-
processes (see also Webster 1992). Sheth et al. (1988) confirms this and argues
further that the mix is a key element in the managerial school of marketing.
According to Grönroos (1994), the main reasons the marketing mix dominated the
marketing literature at that time are as follows:
1) The concept characterised marketing as an easy activity to perform.
2) It allowed the separation of marketing from other organisational activities (for
example, sales), and, hence, encouraged the delegation of marketing activities
to specialists.
3) The elements of the marketing mix can change an organisation’s competitive
situation.
The concept of the marketing mix is not without its critics (see for example,
Booms and Bitner 1982; Kent 1986; Rafiq and Ahmed 1995). Much of the criticism is
based on the point of view that the marketing mix initially included a list of twelve
elements, but was later oversimplified by McCarthy in 1964. Criticism of the model
began when Kent (1986) contended that the marketing mix framework was too
simplistic, and could be misleading as the ultimate strategic model to be applied in
any organisation. Later Webster (1994) raised a different point when he claimed that
the marketing mix did not emphasise the importance of buyer-seller relationships in
industrial marketing.
Möller (2006:4) summarised the criticism of the marketing mix framework
according to four key aspects:
1. The mix does not consider customer behaviour, but instead is internally
oriented.
2. The mix regards customers as passive; it does not allow interaction and cannot
capture relationships.
3. The mix is devoid of theoretical content; it works primarily as a simplistic
device for focusing the attention of management.
4. The mix does not offer help for personification of marketing activities.
Regardless of all the criticism of the marketing mix, Grönroos (1994:4) argued
that the framework remained the dominant marketing paradigm, and had become
Chapter Two – The Development of the Relationship Marketing Concept
-23-
“…the unchallenged basic model of marketing, overpowering previous models and
approaches”. These models included the organic functionalist approach, the systems-
oriented approach and parameter theory9. Moreover, Grönroos (1994) quoted the most
widely accepted definition of marketing, at the time, from the AMA (see section 1.2)
in support of his claim. Grönroos (1994) also noted that a list can never include all the
relevant elements, does not fit every situation, and can become obsolete. Other
marketing academics (for example, Kotler 1986; Baumgartner 1991; Booms and
Bitner 1982) stated, that more Ps should be added to the marketing mix to overcome
all the criticism, either to modify it or to make it applicable to new areas of marketing.
Subsequently, many attempts were made to add to and develop the marketing mix
theory, but with little success. For example, MaGrath (1986) proposed the addition of
three further Ps (personnel, physical facilities and process management), while Kotler
(1986) suggested adding public relations and political power. Baumgartner (1991) put
forward fifteen Ps, Judd (1987) argued that only one element (people) should be
added to the mix, and Goldsmith (1999) suggested expanding the mix to eight Ps
(adding participants, physical evidence, process and personalisation). Booms and
Bitner (1982) proposed the most successful attempt to modify the four Ps model when
they added three additional Ps to fit service marketing (people, physical support and
process). Booms and Bitner’s (1982) additional three Ps are still evident in all modern
academic marketing books (Möller 2006)10
Debates about the marketing mix
.
11 forced marketers to question the
applicability and validity of the “holy” mix (Kent 1986:146). Vargo and Lusch (2004)
argued that attention should be focused on the customer instead of the marketing mix,
because the only purpose of the organisation is to satisfy customer needs. In an effort
to introduce a new marketing paradigm, Vargo and Lusch (2004) labelled the period
from 1980 to 2000 as being characterised by ‘Marketing as a Social and Economic
Process’, a replacement of the pre-1980 marketing mix era. They argued that the
‘dominant logic’12
9 For a full discussion of these theories, see Grönroos (1994). 10 For more detailed discussion see Möller (2006), who provides an up-to-date view of the existing debate surrounding the marketing mix based on a review of academic perspectives from five marketing management sub-disciplines. 11 Appendix Four elaborates on the strength and weakness of the marketing mix. 12 For a detailed discussion on this topic see Lusch and Vargo (2006).
shift in this period was towards marketing as primarily a social and
economic process. They concluded, “...perhaps the central implication of a service-
Chapter Two – The Development of the Relationship Marketing Concept
-24-
centred dominant logic is the general change in perspective” (Vargo and Lusch
2004:12).
In summary, marketing has evolved from the ‘classical’ four Ps framework to
centre around personalised relationships. Consequently, a new concept within
marketing evolved in the last twenty-five years, known as relationship marketing.
2.3 The Emergence of Relationship Marketing13
Berry (1995:236) continued his earlier work when he described relationship
marketing as a “new-old concept”, meaning that “the idea of a business earning the
customers’ favour and loyalty by satisfying their wants and needs was not unknown to
the earliest of merchants”. At the centre of the relationship marketing perspective is
the idea that customers have continuing value over the period they stay in a
relationship with a specific company. Therefore, the focus is on the relationship rather
The concept of relationship marketing was first discussed by Berry in 1983 as
part of a conference paper on service marketing. It was defined as “attracting,
maintaining and in multi-service organisations enhancing customer relationships”
(Berry 1983:25). Berry (1983) recognised that customer acquisition was and would
remain part of the marketer’s responsibilities, but also emphasised that a relationship
view of marketing implied that retention and development were of equal or even
greater importance to the organisation in the long term. However, the concept of
relationship marketing did not attract broader attention until the 1990s (Grönroos and
Ravald 1996).
Marketing researchers (for instance, Berry 1983; Grönroos and Ravald 1996)
argued that the need for a relational paradigm in marketing arose at the point when the
changing business context had constrained organisations to the extent that the only
way they could survive in the marketplace was to cut costs. The shift in focus was
further fuelled by the realisation that retaining customers is more profitable than
allocating huge marketing budgets to go after countless numbers of new customers.
Therefore, keeping existing customers and market share, rather than looking for new
customers and new markets, became paramount (see also Buttle 1996).
13 Appendix Six provides a review of the key propositions if relationship marketing over the past thirty years.
Chapter Two – The Development of the Relationship Marketing Concept
-25-
than on individual transactions. The duration of the exchange is a core element in
distinguishing these two foci: a transactional exchange involves a single, short-term
exchange with a distinct beginning and ending. In contrast, a relational exchange
involves multiple linked exchanges extending over time and usually involves both
economic and social bonds.
Relationship marketing was picked up, during the early 1990s, by marketing
academics and organisations as a response to the shift in business contexts, away from
mass marketing towards personalised or one-to-one marketing by building personal
relationships. This shift caused a great number of organisations to restructure,
establish partnerships and adopt cost reduction measures whilst devising strategies to
retain their customers (Morgan and Hunt 1994; Palmer 1996). As a result, relationship
marketing emerged as a promising way for marketing to enhance social relations with
customers. In other words, relationship marketing was viewed as a method for
organising marketing both within and outside the organisation. The basic proposition
of relationship marketing is that selling organisations should take a long-term view of
customer relationships, to ensure that those customers who are converted are also
retained.
Several authors have embraced relationship marketing as a paradigm shift in
marketing theory14: “… we have to realise that it is a new paradigm, not just a new
model” (Grönroos and Ravald 1996:315; see also, Gummesson 1996; Aijo 1996;
Sheth and Parvatiyar 2001). Although the concept has been largely accepted in
marketing, Backhaus (2000)15
14 This topic will be discussed in detail in section 2.5.1. 15This was also proposed by Lovelock in a keynote speech in the SERVSIG conference (2005, Singapore).
formulated two crucial conditions that must hold for
this new concept to be acknowledged as a paradigm shift in marketing theory. Firstly,
a new paradigm must encompass all the theories and issues in the field; and secondly,
novel methods and tools for theoretical analysis must be provided by the new
paradigm. Relationship marketing is not a useful model in certain market exchanges
(for example, consumer fast goods markets, when transactional strategies are more
suitable), and it draws on pre-existing constructs and solutions (such as customer
loyalty, trust, commitment and service quality) rather than creating new ones.
Chapter Two – The Development of the Relationship Marketing Concept
-26-
The development of relationship marketing theory and practice can be
examined from a number of different perspectives (see Table 2.1), demonstrated by
the existence of three schools of thought (Palmer et al. 2005:313): the Nordic school,
the Industrial Marketing and Purchasing (IMP) Group and the Anglo-Australian
approach (the Cranfield school).
Table 2.1: Comparison between the Major Schools of Relationship Marketing Key component IMP group Nordic school Anglo-Australian
approach Basis Relationship between
firms Service Service/quality/marketing
Time-frame Short- and long-term Long-term Long-term
Organisation
N/A
Functional and cross-functional
Cross-functional, Process-
based
Basis of exchange Product/service,
information, financial, and social
Less sensitive to
price
Perceived value
Product/quality dimension
Technological Interaction quality Function of value and cost of ownership
Measurement
Customer profitability
Quality, value, customer
satisfaction
Customer satisfaction
Customer information
Varies by relationship stage
Individual Customer value and retention
Internal marketing N/A Substantial strategic importance
Integral to the concept
Service Close seller/buyer relations
Integral to product Basis for differentiation
(Source: Adapted from Palmer et al. 2005)16
The Nordic School was originally developed for services marketing research
during the early 1980s, and was characterised by a shift in focus from ideas associated
with traditional marketing concepts (Grönroos 1994; Gummesson 1996). These shifts
included: stressing the importance and relevance of services marketing and industrial
marketing over customers’ goods marketing; a gradual shift away from an emphasis
The Nordic School
16 Palmer et al. based their analysis on the work of: Aijo (1996); Christopher (1996); Christopher et al. (1991); Ford (1994); Grönroos (1994); Kotler (1992); Ravald and Grönroos (1996); Turnbull et al. (1996).
Chapter Two – The Development of the Relationship Marketing Concept
-27-
on goods and services towards an emphasis on customer value; the integration of the
marketing department function with other organisational functions and with general
management; and less emphasis on quantitative research than on qualitative research
(Egan 2004). According to the Nordic school approach, managing services was at the
centre of relationship building and maintenance, although it was supported by other
elements such as the construction of networks, the establishment of strategic alliances,
the development of a customer database or what is referred to as Customer
Relationship Management (CRM)17
The Anglo-Australian approach is based on the work of Christopher, Payne
and Ballantyne, and emphasises the integration of quality management, services
marketing concepts and customer relationships economics (Christopher et al. 2002:4).
From this perspective, relationship marketing entails the following elements: 1)
emphasis on establishing relationships, rather than a simple transactional approach to
marketing, 2) understanding of the economics of customer retention, thereby ensuring
that resources are properly allocated between the two tasks of retaining existing
and the management of relationship-oriented
marketing communications (Grönroos and Ravald 1996).
The IMP Group
During the 1980s, the Industrial Marketing and Purchasing (IMP) group
picked up on relationship marketing as a way to understand business-to-business
markets, emphasising the specificity of the business-to-business sector and the
necessity for new marketing techniques and the development of management tools. In
such markets, transactions are not isolated, but occur instead as part of a continuous
stream of interactions between organisations. The interactions between companies and
many individuals establish the relationship, and the multiple relationships between
buyers, suppliers and other firms aggregate into a network (Palmer et al. 2005:320).
Competitive advantage can be gained from an adequate selection process and
management of network partners.
The Anglo-Australian approach
17 Christopher et al. (1991) embraced CRM as the practical implementation of relationship marketing, containing technological tools in both the hardware and software sides. The aim of CRM is to help managing large number of customers’ data. Therefore, it is beyond the scope of this thesis to discuss the emergence of CRM in relationship marketing theory.
Chapter Two – The Development of the Relationship Marketing Concept
-28-
customers and attracting new ones, 3) a recognition of the important role of internal
marketing in achieving external marketing success, 4) an extension of the principles
of relationship marketing to encompass a range of diverse marketing domains (for
example, customer markets, internal markets, influence markets and referral markets),
5) an acceptance that the traditional marketing mix concept (the four Ps18
) does not
adequately capture all the key elements which must be addressed in building and
sustaining relationships with customers, and 6) a consideration of marketing as a
cross-functional process (Christopher et al. 2002; Palmer et al. 2005).
Although these schools of thought appear to have developed independently
(see Table 2.1), a conflict occurred between the Nordic school and the IMP group
when each claimed to have contributed the most to the foundation of relationship
marketing; hence, each school defined relationship marketing according to its own
perspective. The Anglo-Australian school initially followed the Nordic school in its
ideas, but later attempted to differentiate itself by highlighting its own contribution to
the subject (for example, the other schools’, such as the Nordic school, focus on
quality management in contrast to the IMP’s emphases on customer profitability).
There was also a later debate surrounding the origin of relationship marketing and
whether it was first discussed in Europe or in North America; this will be discussed
later in section 1.5.2.
Berry et al. (1991) argue that relationship marketing is the preferred concept
for services marketing (see also, Grönroos 1994). Sheth and Parvatiyar (1995) built on
Berry et al.’s (1991) work, noting that by engaging in a long-term relationship with
organisations, customers or other companies are able reduce the complexity of buying
situations. In contrast, Palmer (1996) stated that customers often enter into
relationships with marketers in order to increase the resulting number of choices
available to them.
18 See section 2.2.1.
Chapter Two – The Development of the Relationship Marketing Concept
-29-
2.4 Theoretical Foundations of Relationship Marketing
Marketing academics (for example, Grönroos 1994; Möller and Halinen 2000)
debated the theoretical foundations that have contributed to the development of
relationship marketing. For instance, Gummesson (1994) proposed that theoretical
models contributing the most to relationship marketing were: service marketing; the
network approach to industrial marketing; quality management; and, indirectly,
organisational theory. In his discussion about the theoretical foundations of
relationship marketing, Grönroos (1994) argued that the main bases of relationship
marketing are: the marketing of services; the network approach to industrial
marketing; customer relationship economics; and market economies. Möller and
Halinen (2000) have further developed and expanded on both Gummesson (1994) and
Grönroos’s (1994) ideas to include database marketing and direct marketing as an
additional key theories contributing to the development of relationship marketing as
its illustrated in Figure 2.2 and discussed straight after.
Building on the preceding discussion, the modern frameworks which could be
viewed as the forerunners of relationship marketing are: the business marketing and
interaction network approach; the services marketing approach; marketing channels;
(Source: Möller and Halinen 2000:32)
Figure 2.2: The Disciplinary Roots of Relationship Marketing
Chapter Two – The Development of the Relationship Marketing Concept
-30-
and database and direct marketing. These theoretical models of relationship marketing
are discussed below.
2.4.1 Business Marketing and The Interaction Network Approach
The business marketing and interaction network approach (the Nordic school)
is mainly devoted to industrial marketing, and was developed in Sweden at Uppsala
University in the 1960s (Grönroos 1994). Later it was adapted by the IMP Group in
the early 1980s. The model developed by the IMP Group was based on the notion that
various interactions take place in a network of buyers and sellers of industrial goods
and services, including the flow of goods and services and financial and social
exchanges (Johanson and Mattsson 1985). According to Johnson (1986:25) the
network approach represents the “flow of goods and information as well as financial
and social exchanges taking place in the network”. This applies mostly in business-to-
business (B2B) and industrial situations, where such relations lead to shared value
creation for both the seller and the buyer.
2.4.2 The Services Marketing Approach
The services marketing approach can be traced back to the mid 1970s, when
Shostack (1977) published an article in the Journal of Marketing promoting service
marketing as an interesting area of research (Fisk et al. 1994). This approach is based
on the notion that the customer of a service typically interacts with systems, physical
resources and employees of the service provider (see also Håkansson 1982; Zeithaml
et al. 1985; Gummesson 1997)19. Hence, the customer becomes involved in the
production of the service, a process which Langeard and Eiglier (1987) refer to as
‘servuction’20
19 Gummesson (1998) provided a full ‘route’ to achieving total relationship marketing, this is illustrated in appendix Eight. 20 Eiglier and Langeard (1987) were the first to consider incorporating consumption within the services marketing field.
; their work was based on Gummesson’s (1978) research, where he
argued that the marketing approach is focused on facilitating interactions with
customers during their consumption process rather than on the exchange itself. This is
also aligned with Grönroos’s (1978:596) argument that:
Chapter Two – The Development of the Relationship Marketing Concept
-31-
“The consumers are actively taking part in shaping the service offering, i.e. in product development ... the consumer himself can be considered part of the service he buys and consumes.”
Later, Grönroos (2006) argues that the services marketing approach is vital to
organisations because marketers can observe consumption patterns during service
encounters. This is a consequence of the fact that “…given the procedural nature of
services, it follows that the consumption and production of services are at least partly
simultaneous processes, and that the service provider at least partly enters into the
consumption sphere. The production of services is an ‘open system’ for the consumer,
but likewise the consumption of services is an open system for the service provider”
Grönroos (2006:319).
2.4.3 Marketing Channels
Academic research involving marketing channels often involves examining
how actors in these channels behave, and how and why various forms of channels
evolve (Möller and Halinen 2000). The basic prescriptive goal for marketing channels
is to define efficient relational forms between channel members. This goal is
primarily theory-driven, attempting to combine economic, political (or power
dependency) and social aspects (cooperation, trust, commitment, communication, and
conflict behaviour).
It can be argued that marketing channels rely on transaction cost theory,
relational law, social exchange theory, political economy, and power and conflict in
organisational sociology (Anderson and Narus 1990; Williamson 1985). According to
Möller and Halinen (2000) there are three essential points that should be considered
when discussing the marketing channels approach: (1) both economic and political
aspects and their interactions must be considered in examining channel behaviour; (2)
a focal channel/dyad is the recommended unit of analysis; and (3) complex
relationships cannot be understood outside of their context, since the ‘dyadic
behaviour’ and the ‘channel’ are reciprocally interrelated (Möller and Halinen 2000).
Chapter Two – The Development of the Relationship Marketing Concept
-32-
2.4.4 Database Marketing and Direct Marketing
As discussed previously, Möller and Halinen (2000) agreed with Gummesson
(1996) and Grönroos (1994) in their early views about the foundation of relationship
marketing. However, they also argued that database marketing and direct marketing
should be considered when discussing relationship marketing, and came to the
conclusion that there is no universal theory that encompasses all the foundations of
relationship marketing. Consequently, Möller and Halinen’s (2000) main thesis was
that the process of establishing business relationships could not be characterised as a
process of actions and reactions, but rather of interactions that happen at multiple
levels in the organisation. This approach was neither purely management-focused nor
consumer-driven; rather it espoused an inter-organisational orientation in its
description of marketing processes (Möller and Halinen 2000).
2.5 Defining Relationship Marketing
A number of marketing academics have debated how to define relationship
marketing. They often either attempt to provide a holistic definition (see for example,
Gronroos 1994) or to define the term from a specific perspective (see for instance,
Christopher et al. 1991). Harker (1999:13) examined the reasons for these different
approaches, and summarised them as follows:
1. Because it is an emergent perspective, relationship marketing has only had a
relatively short lifetime in which to develop into a fully-formed paradigm.
2. Contributors to the development of relationship marketing theory are
extremely varied, both in terms of socio-political heritage and academic
background.
Table 2.2 (page 34) summarises the key definitions of relationship marketing
mentioned in the literature (adapted from Harker 1999). Some of the key definitions
noted earlier will be discussed in detail in the following section. Moreover, the main
key themes that appear in the definitions of relationship marketing are identified as:
attracting new customers; retaining existing customers; mutual relational exchange;
long-term relationships; added value; strategic focus; and ongoing assessment. These
Chapter Two – The Development of the Relationship Marketing Concept
-33-
themes vary depending on the researchers’ discipline and the time when the definition
was put forward.
The definitions offered in Table 2.2 derive from a number of different research
perspectives, and, hence, they focus on different elements within relationship
marketing. For instance, Grönroos (1994) in his definition proposed a systematic
method for the operation of relationship marketing. Buttle (1996) adapted this
definition with a focus on strategic positioning. Gummesson (1994) focused on long-
term interactivity and profitability between the organisation and its customers.
O’Malley and Tynan (2006:33) stated that “which definition one chooses is likely to
be influenced by the choice of empirical context, the focus of the study (for example,
practical or philosophical), as well as the research stream to which the author
belongs”. This thesis follows the definition proposed by Grönroos (1994) because of
its wide acceptance and the applicability of this definition to the current research
setting.
Despite the diversity in the academic definitions (see Table 2.2), O’Malley
and Tynan (2006:33) identify four emergent issues as follows:
• Relationship marketing refers to commercial relationships between economic
partners, service providers and customers at various levels in the marketing
channel and the broader business environment.
• This recognition results in a focus on the creation, maintenance and
termination of these commercial relationships, in order that parties to the
relationships achieve their objectives (mutual benefit).
• Profit remains an underlying business concern and relational objectives are
achieved through the fulfilment of promises.
• Trust is essential to this process of relationship development and centres upon
the keeping of promises.
Chapter Two – The Development of the Relationship Marketing Concept
-34-
Author Source Definition Author Source Definition
Berry (1983)
Services Marketing
Relationship marketing is
attracting, maintaining and in multi-service organisations
enhancing customer relations.
Gummesson
(1994)
General
Management
Relationship marketing emphasises a long-term interactive relationship between the provider and the customer, and long-term
profitability.
Dwyer et al. (1987)
Buyer/Seller
Longer in duration [than transactional marketing],
reflecting an on-going process.
Grönroos (1994)
Services Marketing
Relationship marketing is to establish,
maintain, and enhance relationships with customers, and other partners, at a profit,
so that the objectives of the parties involved are met.
Christopher et al. (1991)
Services Marketing
Relationship marketing has as its concern the dual focus of
getting and keeping customers.
Buttle (1996)
General Management
Relationship marketing is concerned with
the development and maintenance of mutually beneficial relationships with
strategically significant markets.
Shani and Chalasani (1992:34)
Interaction/Network
An integrated effort to identify, maintain and build up a network with individual consumers and to continuously strengthen the network for the mutual benefit
of both sides, through interactive, individualised and
value-added contacts over a long period of time.
Takala and Uusitalo (1996)
General Management
Customer relations are emphasised. The focus is on the profitable
commercialisation of customer relationships, and the pursuit of individual and organisational objectives. Particular
stress is placed on long-term and enduring relationships with customers.
Table 2.2: Key Relationship Marketing Definitions
Chapter Two – The Development of the Relationship Marketing Concept
-35-
Pathmarajah (1993)
Buyer/Seller
The process whereby the seller and the buyer join in a strong
personal, professional, and mutually profitable
relationship over time.
O’Malley et al. (1997)
Services Marketing
Relationship marketing involves the identification, specification, initiation, maintenance and (where appropriate)
dissolution of long-term relationships with key customers and other parties, through mutual exchange, fulfilment of promises and adherence to relationship norms in
order to satisfy the objectives and enhance the experience of the parties concerned.
Morgan and Hunt
(1994:22)
Buyer/Seller
Relationship marketing refers to all those activities directed
towards establishing, developing and maintaining
successful relational exchanges.
Ballantyne (1997)
General
Management
Buyer/Seller
Relationship marketing is an emergent disciplinary framework for creating,
developing and sustaining exchanges of value between the parties involved,
whereby exchange relationships evolve to provide continuous and stable links in the
supply chain.
Harker (1999:16)
General
Management
Relationship marketing occurs when an organisation engages
in proactively creating, developing and maintaining committed, interactive and profitable exchanges with
selected customers or partners over time.
Sheth and Parvatiyar
(2000)
Buyer/Seller
Relationship marketing is the on-going process of engaging in cooperative and
collaborative activities and programs with immediate and end-user customers to
create or enhance mutual economic value, at a reduced cost.
(Source: adapted from Harker 1999)
Chapter Two – The Development of the Relationship Marketing Concept
-36-
As shown in Table 2.2, there are numerous ways of defining relationship
marketing. In an attempt to provide a general definition to suit all the various
disciplines connected to relationship marketing, Grönroos (1994)21
21 Grönroos’ (1994) definition is one of the most often-quoted definitions of relationship marketing.
proposed a
definition based on his earlier description in 1990. Grönroos (1994:14) viewed
relationship marketing as a process that involves establishing, maintaining and
enhancing long term relationships with customers. His definition implies that the
relationship is a reciprocal process with mutual benefits for both parties over the long
term. Hence, the relationship is a dynamic process with stability as an important goal.
Therefore, a fundamental principle is that organisations engage in relationship
marketing, firstly, because it enhances customer satisfaction, and, secondly, because it
helps to improve other long-term results (Bejou et al. 1998).
Shani and Chalasani’s (1992) definition is almost identical to Grönroos’s
(1990) formulation; both definitions state the importance of individuality in buyer-
seller relationships (Peterson and Smith 1995). In contrast, Morgan and Hunt (1994)
highlighted the importance of relationship trust and commitment in their definition.
However, several marketing academics criticised this description. Roa and Perry
(2000) argued that Morgan and Hunt’s (1994) definition is limited because some of
the details in relationship marketing trees may be lost. For instance, the definition
does not specify the nature or purpose of the relational exchanges. Furthermore,
Peterson (1995) criticised Morgan and Hunt’s (1994) definition and argued that they
might be guilty of an error of commotion in terms of the differences between
‘relational exchanges’ and ‘transactional exchanges’.
Harker (1999) attempted to identify seven conceptual categories extracted
from twenty-six definitions of relationship marketing in the literature. Harker’s (1999)
categories were: creation, development, maintenance, interactive, long-term,
emotional content, and output. This emergent definition emphasised the management
of a number of relationships. However, Harker’s (1999) definition is merely a
summary of all the previous definitions, which makes it very broad in scope. It lacks
any focus on a particular aspect of relationship marketing; in other words, it is theory
driven and cannot easily be implemented.
Chapter Two – The Development of the Relationship Marketing Concept
-37-
Gummesson (1997:270) pointed out that no definition of relationship
marketing can ever be precise and all-inclusive because social phenomena are not in
themselves precise. He argued that definitions of relationship marketing can only be
used as vehicles for thought, as perspectives, or as indicators of essential properties of
a phenomenon. It is difficult, if not impossible, to form a holistic, operational and all-
encompassing definition because the boundaries are limitless rather than clearly
delineated, and perspectives within the research field differ widely. In short, as with
any other social phenomenon, relationship marketing is vague and ambiguous in
nature, making it difficult to define and conceptualise.
2.5.1 Relationship Marketing: A Paradigm Shift
The debate in the marketing literature about whether relationship marketing is
a paradigm shift or an old phenomenon has attracted significant attention from
numerous marketing academics (for instance, Gummesson 1994; Petrof 1998). On the
one hand, several support relationship marketing as a paradigm shift, which occurred
due to a movement from a transaction-based approach towards a relational approach
(Grönroos 1994, 1997; Gummesson 1994, 1996). On the other hand, different
researchers contend that relationship marketing is not a paradigm shift because it has
its roots in the marketing literature (see for example, Egan 2004 and Petrof 1998);
therefore, “relationship marketing is an old idea but a new focus … the forefront of
services marketing practice and academic research” (Berry 1995: 236).
In particular, Gummesson (1994) advanced an argument in support of
relationship marketing as a paradigm shift. He offered a thirty Rs approach within an
operational perspective. The approach emphasised customer retention, the value of
interaction, operational and practical aspects, linkage to organisational structure and
management, and collaboration to compete. However, he also noted that “...the claim
of a paradigm shift in marketing was controversial” (Gummesson 1997:267). As a
result, relationship marketing can be seen as a paradigm shift because it provides a
new foundation for thinking. As a counterbalance Gummesson (1994) also argued that
existing knowledge can be incorporated into a new paradigm, but if a new model
cannot provide its own novel foundation, it is not possible to claim a paradigm shift;
the new approach must be applicable in action. These claims seem partially
contradictory, bearing in mind that relationship marketing is epistemological (i.e. still
Chapter Two – The Development of the Relationship Marketing Concept
-38-
attempting to establish its foundation and create a valid knowledge base) according to
Gummesson (1994).
This thesis takes the stand that relationship marketing is neither a new
paradigm nor a shift in an existing one. This is because that the elements that
contribute to the foundation of the relationship marketing philosophy remain at the
core of marketing thought (Grönroos 1994). These are internally consistent, and unite
harmoniously in order to providing a marketing continuum to propel an organisation
into well-sustained growth. Therefore, to claim a paradigm shift, relationship
marketing would have to be revealed as a phenomenon that does not revolve on an
existing fulcrum (the marketing mix, on which the approach currently rests). From the
perspective of this thesis, such a demonstration of independence is not occurring.
Marketers will continue to introduce new products, set prices, find distribution
channels and promote their products; clearly consideration is being given to customers
in these processes, but this is consistent with the current marketing orientation and not
because of any paradigm shift. Therefore, relationship marketing should not be
viewed as a new paradigm. Instead, it should be viewed as an improved and evolving
paradigm with a new name.
2.5.2 Relationship Marketing as a European or American Model
As mentioned earlier, there is a debate among academics as to whether
relationship marketing is a new paradigm or a new name for an old phenomenon.
Takala and Uusitalo (1996:45) attributed the emergence of relationship marketing
solely to northern European researchers, claiming:
“Researchers in northern Europe and within the fields of industrial and service marketing found in the 1970s that a new concept of marketing was needed, one which stresses the relationship between the seller and the customer.”
Grönroos (1994) credited the emergence of relationship marketing to
European and Australian academics, but he also gave some of the credit to North
American researchers and the IMP Group in Europe. Gummesson (1997) suggests
that both services and quality are viable areas of research and practice in North
America, but pointing out that this has not yet occurred with the network approach to
industrial marketing (see also, Palmer et al. 2005). What is evident in this perspective
Chapter Two – The Development of the Relationship Marketing Concept
-39-
is the fact that Europe is leading the development of the theory, but research and
literature are by no means exclusive to it (see for example, Håkansson 1982; Johanson
and Mattsson 1985).
This argument probably stems from the contributions made by European
researchers, shown in the interpretations of the history of services marketing
presented by Berry and Parasuraman (1993) and Fisk et al. (1994). This interpretation
contradicted the previously-accepted scenario wherein US thinking about marketing
was better known than its European equivalent (Gummesson 1997; Palmer et al.
2005).
It can be argued that current thinking about relationship marketing has become
universal through effective distribution via journals, publishers, associations, and
conferences. Relationship marketing and its roots are seen as European models
because European thinking in these fields seems to have been disseminated more
effectively than that emanating from the United States.
From an analytical point of view, the idea that relationship marketing is an old
phenomenon is not specifically European or American. It may be a new current
direction in the literature, but it is accepted that the practices associated with
relationship marketing have long been applied in many countries, as well as by
previous generations (see for instance, Grönroos 1994). However, it is acknowledged
that Europeans did a great deal of work to create the term relationship marketing as it
is known today – but this doesn’t make the phenomenon, or for that matter the model,
exclusively a European creation.
Chapter Two – The Development of the Relationship Marketing Concept
-40-
2.6 The Core Elements of Relationship Marketing
The relationship marketing concept, as discussed earlier, is based on the
creation of a mutual long-term relationship between the engaged parties (see for
example, Grönroos 1994, 2000; Shani and Chalasani 1992; Palmer et al. 2005) in
conjunction with certain concepts designed to fulfil the needs and wants of the parties
engaged in the relationship. These features, such as the concept of trust, are discussed
next.
2.6.1 Trust22
Trust has been defined as “a willingness to rely on an exchange partner in
whom one has confidence” (Moorman et al. 1993:3), and it is viewed as an essential
ingredient for successful relationships (Berry 1995; Dwyer et al. 1987; Morgan and
Hunt 1994). The development of trust occurs when individuals have “confidence in an
exchange partner’s reliability and integrity” (Morgan and Hunt 1994:23). However,
this development does not take place over a short period; time is needed to establish
trust (Garbarino et al. 1991; Morgan and Hunt 1994; Garbarino and Johnson 1999).
Therefore, organisations should expect that it might take a long time to capture
customers’ trust (Malecki and Tootle 1996; Macintosh 2002).
The concept of trust has been discussed from many different perspectives.
Economists tend to view trust as either calculative or institutional (Korczynski 2000),
psychologists tend to focus on the attributes of the trustors and the trustees (Rotter
1980), and sociologists often find trust in the social bonds resulting from relationships
between people (Wicks et al. 1999). Morgan and Hunt’s (1994) work is considered to
be the foundation of trust and commitment research within marketing; they proposed
some key mediating variables for relationship marketing (see Figure 2.3).
22 The concept of trust and will be discussed in greater detail in the following chapter.
Chapter Two – The Development of the Relationship Marketing Concept
-41-
There are many different levels of trust. Perhaps the most common types of
trust are: multi-level trust (between individuals, groups, firms and so on); trust within
and between organisations; and new or emerging trust. Nevertheless, the risk involved
in any exchange or relationship between organisations and customers depends mainly
on the form and depth of the relationship between the service provider and its
customers (Sheppard and Sherman 1998; Macintosh 2002). Furthermore, trust has
(Source: Morgan and Hunt 1994:22)
Figure 2.3: The Key Mediating Variables of Relationship Marketing
Chapter Two – The Development of the Relationship Marketing Concept
-42-
been recognised as a means for reducing operational problems, both between
employees and managers and across organisations (Rousseau et al. 1998).
Trust is the key element binding relationship marketing together leading to
relationship commitments. Therefore, the concept will be discussed at greater length
in the next chapter.
2.6.2 Commitment
Commitment is seen as a central element in creating an effective relationship
because it leads to important outcomes such as decreased turnover (Porter et al. 1974;
see Figure 2.3), higher motivation (Farrell and Rusbult 1981), and increased
organisational citizenship behaviours (Williams and Anderson 1991). However, it also
results from organisational support (Eisenberger et al. 1990). Morgan and Hunt
(1994:23) defined commitment as “an exchange partner believing that an ongoing
relationship with another is so important as to warrant maximum efforts at
maintaining it.” However, it has been noted that commitment in a relational exchange
is an important element in relationship marketing. Moreover, commitment between
two exchange parties should be based on trust as an antecedent. The seller should
commit to deliver a high level of quality and to stabilise the relationship in the long
run, and the customer should commit to maintaining and value such a relationship
(Morgan and Hunt 1994). In other words, the future prospect of a buyer-seller
relationship is based on commitment.
Summary
The shift from transactional-based marketing to social and relational
marketing, as discussed earlier, illustrated the significant role of the relationship
marketing concept in the modern marketing literature. Nevertheless, much of the
published work in the area of relationship marketing has been criticised because it
gave the impression that the model is suitable for many marketing contexts, even
though little empirical evidence has been provided.
The next chapter explores the concept of trust and its development across
various areas of the academic literature. It will go on to discuss the important of this
concept within marketing research.
Chapter Three – The Concept of Trust
-43-
CHAPTER THREE: THE CONCEPT OF
TRUST
3.1 Introduction
The concept of trust has been examined in various contexts over the years; for
instance, in industrial buyer-seller relationships (Doney and Cannon 1997),
distribution channels (Dwyer et al. 1997), strategic alliances (Das 1998) and
marketing research (Moorman et al. 1992). Within these differing contexts academics
have examined trust from several perspectives. For example, psychologists defined
trust as a foundation for social exchange between interacting parties to examine why
and how people trust within a social context (Spekman 1988), while economists
conceptualised trust in terms of economic theories such as the prisoner’s dilemma and
game theory (Axelrod 1984). General research on trust also viewed the concept as a
foundation for interpersonal relationships (Rempel et al. 1985; Svensson 2001) and as
the basis for stability in organisational operations (Ennew and Sekhon 2003).
The previous chapter investigated the concept of relationship marketing and
how it came to prominence. This chapter will discuss the concept of trust – its
components, forms and outcomes. The discussion will also highlight different
perspectives on trust, and will conclude that trust is vital for creating long-term
relationships. Finally, it will provide a primary base for proposing a conceptual
framework for trust.
Chapter Three – The Concept of Trust
-44-
3.2 The Concept of Trust
It is widely agreed that trust is an elusive concept that means different things
to different people, with the consequence that it is relevant to several academic
disciplines (Nooteboom 2002; Lane and Bachmann 1998). For instance, economists
tend to focus on the cost or cheat factor in developing trust, and also claim that trust
may be linked to precise situations in which the perceived risk depends on the
behaviour of the different parties involved (Korczynski 1994). Rousseau et al.
(1998:395) stated, from a psychologist’s point of view, that “…the meaning of trust,
as the willingness to be vulnerable under conditions of risk and interdependence, is a
psychological state that researchers in various disciplines interpret in terms of
‘perceived probabilities’, ‘confidence and positive expectations”.
Because of this variation in perspectives among trust researchers, it can be
argued that, trust as a concept, is complex and has several meanings, and
consequently there is no universal definition of the concept (Rousseau et al. 1998).
However, there are several characteristics that span the different discipline domains;
Mayer et al. (1995) identified five key issues that summarise the difficulties in
conceptualising trust:
• The complexity of defining the concept of trust.
• Confusion in agreeing on the antecedents and corollaries of the concept of
trust.
• Vagueness in expressing the correlation between trust and the associated risk.
• Confusion in levels of analysis due to the lack of specificity of trust referents.
• Failure to provide an accurate identification of the attributes and behaviours of
both parties (the trustor and trustee), to which they behave accordingly.
From the above, it is clear that that the concept of trust is difficult to explore,
and as a result it is hard to measure. Nevertheless, several studies have managed
successfully to provide reliable qualitative and empirical evidence on how trust is
perceived in the various domains (Young 2006); these works will be discussed in
greater detail later in this chapter, as will the different definitions, typologies and
classifications of trust.
Chapter Three – The Concept of Trust
-45-
3.3 The Meaning of Trust
The literature on trust demonstrates that the concept of trust is diffuse.
According to Williamson (1993:453), “trust is a term with many meanings”, while
White (1992; 174) sees it as “a term for a clustering of perceptions”. The compound
nature of the concept of trust and the vagueness within the concept is well stated by
Porter et al. (1974:497), who propose:
“Trust … tends to be somewhat like a combination of the weather and motherhood; it is widely talked about, and it is widely assumed to be good for organizations. When it comes to specifying just what it means in an organisational context, however, vagueness creeps in.”1
From the above quotation it is clear that the meaning of trust differs in the
literature according to the perspective from which it is originally defined. More
specifically, most of the marketing literature on trust conceptualises and examines two
main views; these are the calculative view (following an economic model) and the
social and affective view
2
1 Porter et al. (1975) clearly recognised the vagueness within the concept. However, trust does not essentially have to be well defined within the field of marketing. 2 This view is a mixture of the psychological and sociological schools of thought.
(Huemer 1998). These two approaches can be seen as a
platform for conceptualising and understanding trust within organisations.
Trust is often viewed as a key variable in the foundation of a relationship, and
is an antecedent to the corollary of commitment, defined as “confidence in an
exchange partner’s reliability and integrity” (Morgan and Hunt 1994:23). Trust is also
defined as “the perceived credibility and benevolence of a target of trust” according to
Doney and Cannon (1997:36). Another definition of trust has viewed it as “the
confidence that the other party to an exchange will not exploit one’s vulnerabilities”
(Korczynski 2000:4). A final alternative definition can be expressed as “… [the]
willingness of a party to be vulnerable to the actions of another party based on the
expectation that the other will perform (or not perform) a particular action important
to the trustor, irrespective of the ability to monitor or control that other party” (Mayer
et al. 1995:712).
Chapter Three – The Concept of Trust
-46-
Therefore, capturing a universal definition is by no means easy because of the
multiplicity of different perspectives. In addition, it must be borne in mind that
definitions of trust in the literature tend to reflect the paradigms of the particular
academic field of the researcher (Lewicki and Bunker 1995). For instance, while
sociologists tend to see trust as structural in nature, several psychologists have defined
trust as a personal attribute (Rotter 1967). Social psychologists are more liable to
define trust as an interpersonal phenomenon (Svensson 2001), whereas economists
tend to view trust more as a rational choice attitude (Williamson 1993).
From the discussion so far it is apparent that a universal definition of trust
does not exist, and that there are competing definitions. These are summed up in the
typology in Table 3.1.
Chapter Three – The Concept of Trust
-47-
Table 3.1: Key Definitions of Trust
Author Source Definition Author Source
Lewis and Weigert (1985)
General management Social exchange
Trust exists in a social system insofar as the members of that system act according to and are secure in the expected futures constituted by the presence of each other or their symbolic representations.
Zaheer and
Venkatraman (1995:375)
General management Social exchange
Trust reflects the extent to which negotiations are fair and commitments are upheld.
Koller (1988:266)
General management Social exchange
A person’s expectation that an interaction partner is able and willing to behave promotively towards the person, even when the interaction partner is free to choose among alternative behaviours that could lead to negative consequences for the person. The degree of trust can be said to be higher the stronger the individual holds this expectation.
Mayer et al. (1995:712)
General management Psychology
Willingness of one party to be vulnerable to the actions of another party based on the expectation that the other will perform a particular action important to the trustor, irrespective of the ability to monitor or control that other party.
Anderson and Weitz
(1989: 312)
General management Economic
One party’s belief that its needs will be fulfilled in the future by actions undertaken by the other party.
Doney and
Cannon (1997:36)
Marketing Buyer/seller
The perceived credibility and benevolence of a target of trust.
Anderson and Narus (1990: 45)
General management Economic
The organisation’s belief that another company will perform actions that will result in positive outcomes for the organisation, as well as not take unexpected actions that would result in negative outcomes for the organisation.
Rousseau et al. (1998: 395)
General management Psychology
Trust is a psychological state comprising the intention to accept vulnerability based upon positive expectations of the intentions or behaviour of another.
Chapter Three – The Concept of Trust
-48-
Crosby et
al. (1990:70)
Marketing Economic
A confident belief that the salesperson can be relied upon to behave in such a manner that the long-term interest of the customer will be served.
Lewicki et al. (1998: 439)
General management Psychology
Confident positive expectations regarding another’s conduct.
Moorman
et al. (1992:23)
Marketing Social exchange
Willingness to rely on an exchange partner in whom one has confidence.
Sztompka (1999:25)
General management Social exchange
A bet about the future contingent actions of others.
Morgan and
Hunt (1994:23)
Marketing Buyer/seller Social exchange
Trust exists when one party has confidence in an exchange partner’s reliability and integrity.
Korczynski (2000:41)
General management Social exchange
Trust is the confidence that the other party to an exchange will not exploit one’s vulnerability.
Mohr and Spekman
(1994:137)
General management Psychology
The belief that a party’s word is reliable and that a party will fulfil its obligation in an exchange.
Falcone and Castelfranchi (2001:236)
General management Psychology
Trust … is a mental state, a complex mental attitude of an agent x towards another agent y about the behaviour/action relevant for the result (goal) g. Trust is the mental counterpart of delegation.
McAllister (1995:25)
General management Social exchange
The extent to which a person is confident in, and willing to act on the basis of words, actions, and decisions of another.
McKnight and
Chervany (2000:1)
General management Social exchange
The subjective assessment of one party that another party will perform a particular transaction according to his or her confident expectations in an environment characterised by uncertainty.
Chapter Three – The Concept of Trust
-49-
As can be noted from Table 3.1, Rousseau et al.’s (1998) definition of trust is
one of the most quoted definitions in the psychology literature. Their definition lists
two key dimensions of trust, namely vagueness and expectation. However, their
definition does not consider the risk in a relationship incurred by creating trust. In an
earlier work, Lewis and Weigert’s (1985) view of trust presents the idea that members
of a system have expectations of their relationships within that system. Nonetheless,
this definition (like that of Rousseau et al. 1998) seems to assume that customer’s
expectations are always positive.
Morgan and Hunt’s (1994) definition implies a significant degree of reliance
by all related parties. This involves a mutual relationship between the trustor and the
trustee. Mayer et al, (1995) focus on vulnerability, willingness and expectations as the
foundation for trust, with vulnerability representing the fact that losses can increase
when trust is misplaced.
From a marketing perspective, Doney and Cannon (1997) provide a much
simpler definition of trust; they focus on two main dimensions, credibility and
benevolence. Although this definition is intended to describe trust, in fact it implies
trustworthiness rather than trust (a more detailed discussion will follow later on in this
chapter). McKnight and Chervany (2000:1) attempted to build on all the previous
work; their definition captures two significant characteristics of trust. These are the
confident expectation of a possibility of mutual benefits, and the uncertain
environment in which trust can take place.
To put it briefly, as can be seen from Table 3.1 the majority of the definitions
of trust contain three common themes, which Milne and Boza (1999) summarised as:
1. A level of interdependence between the trustor and the trustee.
2. Trust as a way to cope with risk or uncertainty.
3. A belief or expectation that the other party in a relationship will not take
advantage of the vulnerability that arises when the risk of trust is accepted.
The first common theme is that trust implies interdependence, based on the
analysis that the need to trust arises from the existence of social relationships, since
individuals would not have a need to trust without such relationships (Lewis and
Weigert 1985). Expectations about another person’s trustworthiness only become
Chapter Three – The Concept of Trust
-50-
relevant when an individual’s activities depend on the fact that the other individual
cooperates or acts in a certain way (Lane and Bachmann 1998).
Trust can be related to relationship risk in situations where individuals stand to
lose something, for example, reputation, information and so on (Mayer et al. 1995;
Sztompka 1999). Therefore, trust is often associated with risk, but it is not treated
from the same perspective (Mayer et al. 1995). Consequently, trust can be seen in
terms of risk, since it relies on the readiness of the engaged parties to undertake risk
(Luhmann 1988).
The third theme implies that a corollary of risk and uncertainty is vulnerability
(Korczynski 2000). When forming trust in a new relationship, individuals expose
themselves to the risk that the other party will fail to match with their preconceived
expectations, which as a consequence implies vulnerability (Mayer et al. 1995). In
this definition the level of each feature varies depending on the exchange partner’s
characteristics, and, therefore, the formation of trust and the perception of the value of
that trust changes accordingly (Korczynski 1994).
It can be argued that trust cannot be equated with confidence, since confidence
applies to a situation in which there is no necessity for an individual to consider
alternatives. Also, the definition implies that in situations of trust, risks are recognised
and a conscious decision is made to accept these risks (McAllister 1995). Hence, in
situations where confidence has been established, individuals will not consider
alternative options and will simply accept all risks (Deutsch 1960; Luhmann 1988).
Owing to the different schools of thought within trust research, the concept of
trust has been defined in various ways (Corritore et al. 2003). However, what is
common across the various definitions of trust is that they all view trust in terms of a
trustor’s psychological state, such as confidence (Morgan and Hunt 1994) or positive
expectations (Rousseau et al. 1998). The definitions generally contain two involved
parties, the trustor and trustee (Crosby et al. 1990), focusing on the fact that trust is
necessary to reduce risk (McKnight and Chervany 2000) and minimise any
exploitation in the ongoing relationship (Korczynski 2000).
Chapter Three – The Concept of Trust
-51-
This thesis takes the definition of trust provided by Mayer et al. (1995) as a
working definition, because it has been widely accepted among researchers in the area
of relationship marketing and trust (for example, Castaldo 2003).
3.4 Different Perspectives on Trust
Various disciplines view the concept of trust in a different manner; for
example, philosophy (Williams 1988), political science (Dunn 1988) and social
science (Hart et al. 1986). McAllister (1995) distinguished between two main types of
trust; cognitive trust and affective trust (or calculative trust and non-calculative trust).
These two types can be seen as the basis of a conceptualisation of trust, and will be
briefly described below.
Cognitive based trust is a belief, a feeling, or an expectation about an
exchange partner’s trustworthiness that is influenced by the partner’s expertise,
reliability or dependability (McAllister 1995; Moorman et al. 1992; Rempel et al.
1985). Cognitive trust is similar to economic theory, in that the customer can calculate
the benefits arising from their interaction with an organisation based on reputation and
cost; within this perspective opportunism is often associated with trust, in terms of
maximising opportunism through calculating the result of an exchange (Korczynski
2000). Accordingly, the cognitive or calculative perception of trust emphasises the
associated risk in addition to the assumption that the engaged parties are able to
calculate and predict the potential consequences (profits or losses) of a certain
exchange. This form of trust develops via the development of familiarity where the
customer’s reliability and dependability and expectations concerning the service
provider have been demonstrated (Huemer 1998).
Affect based trust is a behavioural intent or a behaviour that expresses the
dependable nature of the engaged parties, in a relationship where one party relies on
the other and the dependence includes a level of vulnerability and uncertainty
(Moorman et al. 1993). The trustor makes an emotional investment in the
relationship, expresses actual care and concern for the other party, and believes that
those sentiments are reciprocated (McAllister 1995). Affective trust can be seen as the
opposite of cognitive trust, since it is based on emotional ties in relationships and is
often built on elements such as concern and benevolence. Additionally, affective trust
Chapter Three – The Concept of Trust
-52-
focuses on the motives and intentions of the exchange partner (Ganesan 1994;
Spekman 1988), instead of on their precise behaviour (Huemer 1998).
In summary, McAllister (1995) distinguishes between trust’s cognitive and
emotional aspects. His study found evidence for a clear distinction between affective
and cognitive trust, both in terms of factor separation, and in terms of distinct
relationships with other concepts. McAllister (1995:49) concluded that “affect-based
trust and cognition-based trust represent distinct forms of interpersonal trust.”
From the standpoint of this thesis, McAllister’s (1995) classification does not
encompass the entire concept of trust, and is seen as an over-simplification of the
concept. The reason for this is that the concept of trust has been explored in detail in
various other domains (economics, sociology and psychology); each viewed trust
according to related theories within that particular field. These differing views of trust
are further elaborated below.
3.4.1 Calculative (Economic) View of Trust
The calculative view of trust reflects an economic perspective (Huemer 1998).
In this viewpoint, individuals are seen as rational actors who are keen to gain maximal
utility from the other party (Huemer 1998). Within this perspective, opportunism is
often associated with trust, in terms of maximising opportunism by calculating the
consequences of the exchange (Korczynski 2000; Williams 1988). Therefore, the
calculative view highlights the risk associated with trust, in addition to the assumption
that the engaged parties are able to calculate and predict the potential consequences
(whether profit or loss) of a certain exchange.
Williamson (1993) suggested three essential sources of trust in the business
world: familiarity, calculative-ness and values3
• Knowledge based trust: trust takes place when enough information for
predicting others’ behaviour is available.
. A clearer identification was
developed by McAllister (1995) to conceptualise the different forms of trust within
the calculative view (McAllister 1995). These are:
3 Williamson’s (1993) work will be discussed in more detail below.
Chapter Three – The Concept of Trust
-53-
• Process based trust: this form of trust, discussed by Zucker (1986), relies on a
common base of knowledge, which means that common background
expectations are required for its production. Nonetheless, “…the development
of trust relies on the formation of a trustor’s expectations about the motives
and behaviours of a trustee, because of the broad nature of trust and its varied
conceptual roots” (Doney et al. 1998:37).
• Fragile trust: fragile trust is based on an individual’s judgments about the
probability of the other party’s actions. Ring (1996) indicates that this type of
trust is similar to McAllister’s (1995) definition of cognitive based trust,
which is also related to predictability. It is argued that fragile trust allows
economic actors to deal with each other in guarded ways, relying on formal or
contractual means and on institutional safeguards.
The classical theory of transaction cost economics shows that decisions about
whether trust is produced in a relationship rely on ‘asset specificity’4
Other economic practitioners, such as Dasgupta (1988), argued that the
creation of trust takes place in a calculative manner, with the focus on the outcome of
the relationship. According to the neoclassical approach, a perfectly competitive
market is populated by large numbers of anonymous, sensitive buyers and sellers who
meet for an instant to exchange standardised goods. All actors in a perfectly
competitive market seek to maximise their own welfare, but the model does not allow
; this notion
emerged in conjunction with the idea of reducing the costly effects of opportunistic
behaviour (Williamson 1993).
Moreover, Williamson (1993) contended that if an appropriate ‘contractual
safeguard’ exists then economic exchange does not depend on trust. Williamson
(1993:467) provided examples of a few situational antecedents for economic
(calculative) trust: the engaged parties (1) are aware of the range of possible outcomes
and their associated probabilities; (2) take cost-effective actions to mitigate hazards
and enhance benefits; (3) proceed with the transaction only if expected net gains can
be projected; and (4) if X can complete the transaction with any of several Ys, the
transaction is assigned to that Y for which the largest net gain can be projected.
4 Williamson (1985:55) defines asset specificity as “durable investments that are undertaken in support of particular transactions, the opportunity cost of which investments is much lower in best alternative uses or by alternative users should the original transaction be prematurely terminated”.
Chapter Three – The Concept of Trust
-54-
for the pursuit of self-interest through acts of deceit or the withholding of relevant
information. Even if agents are permitted to have a desire to behave dishonestly, the
assumption of perfect competition includes a secondary assumption of perfect
information, so that the opportunity for dishonest behaviour never arises. Under the
conditions of a perfectly competitive market, there is no need to trust or distrust
(Platteau 1994).
Based on Platteau’s (1994) discussion, the concept of trust cannot occur within
and as a consequence of the model of perfect competition. However, in reality
markets are not perfect and opportunities for untruthful behaviour flourish. Without a
conceptual framework that includes these features of trust, as this thesis argues, the
concept stays beyond the scope of classic economic examination (Platteau 1994).
An alternative view to the neoclassical analyses of the coordination of
economic activity is provided by transaction cost theory. This school of thought forms
the cornerstone of ‘new institutional economics’, and is most notably associated with
Williamson (1985). Unlike neoclassical economists, Williamson (1993) does not treat
transactions as costless, but recognises that costs are incurred in “obtaining relevant
information, the cost of bargaining and making decisions, and finally, the costs of
policing and enforcing contracts” (Hodgson 1988:180).
Under certain circumstances transaction costs may appear insignificant, but in
others they balance the advantages of the exchange. Moreover, economic theory
introduced several economic models, for instance, the prisoner’s dilemma model and
game theory5
The concept of the prisoner’s dilemma, according to Macy and Skvoretz
(1998), is based on the notion that in everyday life the prisoner’s dilemma exists in
the context of communication between individuals. They give the example of
marriage, suggesting that partners may be living in a constant prisoner’s dilemma
situation if trust doesn’t exist between them to a sufficient degree. Otherwise the
option of advantageous alternatives will constantly exist.
, in relation to how trust is formed in situations calling for cooperation or
competition (Macy and Skvoretz 1998).
5 Game theory; according to (Axelrod 1984), is based on the counter-factual assumption that actors’ behaviour is exclusively determined by calculation. This is not only an extremely simplified view of the socio-economic world, but no matter whether the model is meant to function as an heuristic device or an empirically testable hypothesis, the assumption places game theory models way beyond the scope of realistic empirical research perspectives.
Chapter Three – The Concept of Trust
-55-
Trust in exchange relationships has been conceptualised to fit with different
economic perspectives because it has been shown to be a significant antecedent to
effective inter-organisational collaboration (Sake 1991; Smith et al. 1995).
Furthermore, trust within an economic context often has the following results:
1. Lower transaction costs and the potential for greater flexibility to react to
changing market settings (Dyer 1997).
2. Superior information systems that enhance harmonisation within organisations
and reduce inefficiencies (Dyer 1997).
3. Facilitation of investments in transactions or relation-specific assets that
enhance productivity.
In short, economists are inclined to view trust as a rational choice mechanism
(see Williamson 1993). However, the economic theory of trust has been criticised
both from a conceptual perspective and from a methodological viewpoint; for
instance, some of the economic theories were criticised as being methodologically
questionable, since it is only in the repetitive version of the pure prisoner’s dilemma
that trust becomes relevant. The level of trust during the period of the relationship
does not remain constant. Clark and Sefton (2001) argue that the level of trust is
negatively influenced by any negative experience within the transaction; this is called
distrust, as discussed in section 3.5.
3.4.2 Social View of Trust
The social view of trust emphasises social bonds, individuals’ interactions and
their willingness to maintain respectful relationships (Huemer 1998; Rousseau et al.
1998; Jackson 1985). According to the social view, human relationships are expressed
from the perspective that the feelings that individuals hold towards trust are social in
nature, and therefore, the purely calculative conceptualisation of trust is an inadequate
explanation (Spekman 1988; Zucker 1986). According to McAllister (1995) there are
several types of trust that are defined in terms of the social view of the concept; these
types can be expressed as:
Affect-based trust; trust that is built on interpersonal care and concern for the
welfare of individuals, rather than on self-interest. McAllister (1995) proposes that
this type of trust is based on emotional bonds between individuals; consequently,
Chapter Three – The Concept of Trust
-56-
relationships between the people concerned as well as communication are of great
importance for the development of affect-based trust.
Characteristic-based trust; called ‘free trust’ by Zucker (1986), implying that
this form of trust cannot be invested in or deliberately formed. It depends on the social
similarities or cultural agreements between the engaged parties; consequently,
information about social similarity is necessary for characteristic-based trust
(McAllister 1995).
Institutional-based trust; this type of trust does not depend on personal
characteristics (such as cultural values) or on past history exchange, but instead it
relies on formal social structures, individual or organisation-specific attributes
(Zucker 1986).
Sako (1992) also identified three forms of trust within a social context:
• Contractual trust;
• Competence trust; and,
• Goodwill trust.
Contractual trust is the belief that the engaged parties in a relationship will
adhere to universal ethical standards (i.e. maintain promises). However, these beliefs
emerge from basic notions concerning human nature, social organisation and business
relationships. Competence trust is the belief that the engaged parties in a relationship
will be capable of effectively delivering the contracts and promises they have made.
The third form of trust identified by Sako (1992) is goodwill trust which is the belief
that the engaged parties in a relationship will perform towards a common interest in
addition to their official promises, and will avoid opportunism. Moreover, the three
forms of trust support each other and may stabilise or destabilise each other.
3.4.3 Psychological View of Trust
From a psychological perspective trust is defined mainly in terms of
‘interpersonal trust’. It is stressed that in everyday situations one relies on
expectations related to the behaviour of another person, on his or her promises and
willingness to cooperate. Thus, trust becomes a central precondition for the formation
of a positive interpersonal relationship (Rotter 1967; Zand 1972). The complex nature
Chapter Three – The Concept of Trust
-57-
of trust has been explored by psychologists, and different characteristics of trust have
been associated with the concept; for instance, aspects of uncertainty, as well as the
risk of disappointment from existing relationships (for example, see Rotter 1967).
Empirical research has showed that psychological violation is relatively
common in relationships and that it is associated with various negative consequences
for the relationship, such as a decrease in perceived obligations and reduced
commitment and satisfaction between the engaged parties (Rousseau et al. 1998).
Psychological contract theorists assert that the effect of disrupting a
psychological contract involves more than unmet expectations, because it involves not
only a loss of something expected but also erosion of trust and of the foundation of
the relationship between the two parties (Rousseau et al. 1998). As Rousseau et al.
(1998:396) stated, “… the intensity of the reaction [to violation] is directly
attributable not only to unmet expectations of specific rewards or benefits, but also to
more general beliefs about respect of persons, codes of conduct and other patterns of
behaviour associated with relationships involving trust.”
3.4.4 Sociological View of Trust
The sociological concept of trust is illustrated by the notion that in terms of
one party’s manner toward another party in a precise setting (Mayer et al. 1995); trust
is situated within a complex social organism (Andaleeb 1996; Spekman 1988;
McAllister 1995; McKnight et al. 1998; Yamagishi and Yamagishi 1994). The trustor
is engaged towards a trustee, and the entire relationship takes place within a setting
and/or a context (Clark and Payne 1997).
Socially defined trust relationships can exist not only between individuals, but
also between individuals and organisations. This type of trust, which exceeds the
boundaries of purely personal trust relationships, will expand to become ‘system trust’
if the engaged parties include the entirety (or at least the central elements) of an
institutional environment. In a social sense, trust has been considered predominantly
in terms of a central belief about the motives or intent of another party (Andaleeb
1996). Further, as a social construct, trust relies on relationships and contracts which
act to influence each party’s behaviour towards the other (Spekman 1988).
Chapter Three – The Concept of Trust
-58-
Several authors (for example, Grönroos 1985 and Mayer et al. 1995) have
emphasised the importance of personal relationships in the process of creating trust
(see also Granovetter 1985; McAllister 1995). Furthermore, Granovetter (1985) noted
that the most fundamental factor in the creation of trust is not institutional
arrangements, but instead is the social relations between the engaged parties.
Granovetter (1985) continued this debate and proposed that individuals are concerned
with the level of confidence that the trustee is going to carry out their part of the
agreement, which is linked to whether former contacts between the parties have been
satisfactory.
Lorenz (1988) emphasised the importance of personal relationships when he
proposed that trust can be created intentionally by focusing on long-term relationships
rather than short-term ones. This can be achieved by mutual exchange between the
engaged parties.
It is widely accepted that personal relationships do not necessarily require
friendship when forming trust. Lorenz (1988) stresses the need for personal contacts
as well as geographical proximity for an individual to determine whether to trust or
not. In addition, Lorenz (1988) postulates that an individual cannot exclusively
depend on reputation to determine trustworthiness, but that time and their own
experience are crucial elements. The presence of a personal relationship is more
important in relation to affect-based trust (McAllister 1995). However, affect-based
trust depends on the emotional bonds between the engaged parties, and is
consequently most likely to develop and deepen through personal relationships that
occur over a long period of time (McAllister 1995). Hence, mutual knowledge and the
sharing of information are crucial for the development of this form of trust.
The sociological view of trust has been criticised by several academics. For
instance, Korczynski (2000) argues that politeness (a dimension that is often
discussed within sociological research on trust) will not be relevant if the engaged
parties do not meet because they cannot show their politeness to each other.
Consequently, there is no opportunity for the development of trust based on personal
relations. Furthermore, it is not clear how far all the situational determinants, even in
primary relationships, can be satisfactorily replicated in these scenarios (Korczynski
2000).
Chapter Three – The Concept of Trust
-59-
Rempel et al. (1985) proposed a conceptual framework of trust dynamics in
romantic relationships. They explored the issue of trust production by proposing a
model consisting of three components that reflect increasing levels of behavioural
attributes. The ingredients of the model are predictability6
6 See also: Coleman 1990; Gambetta 1988; Deutsch 1973.
, dependability and faith.
These dimensions are discussed in detail below.
Predictability: this can be perceived as a precise behaviour (the formation of
general expectations) from one party towards another party’s potential behaviour,
based on conclusions drawn from experience and past history. It refers to the other
party’s consistency of behaviour (Doney et al. 1998). This consistency can be affected
by the social setting in which the relationship takes place (for example, in
relationships between different social classes). Accordingly, the stability of the
psychosocial setting should also be considered when assessing another party’s
predictability (McKnight et al. 2002). Rempel et al. (1985) propose that the
judgement about another party’s predictability is facilitated if one party possesses
specific information about the other party’s potential or past behaviour, for instance,
in terms of reinforcements and restraints.
Dependability: Rempel et al.’s (1985) framework of trust states that with time
there is a shift from assessing another party’s specific behaviours to evaluating the
qualities and characteristics attributed to that party. As they put it, “… trust is placed
in a person, not his or her specific actions”. Thus, dependability refers to the partner’s
moral integrity, encompassing factors such as benevolence, reliability, honesty, and
concern with providing expected rewards (Blois 1999; Mayer et al. 1995; Larzelere
and Huston 1980). The evaluation of such personal qualities will largely be informed
by experiences involving risk and personal vulnerability. The dependability
component is clearly related to predictability, although it is more involved with a
“sub-class of behaviours that involve personal vulnerability and conflicts of interests”
(Rempel et al. 1985:95). As with predictability, dependability also assumes that a
relationship lasts long enough to allow for a detailed analysis of the other party’s
trustworthiness. When faced with novel situations within an existing relationship, or
when starting a new relationship, one must consider the third component, faith.
Chapter Three – The Concept of Trust
-60-
Faith: in the context of trust, faith refers to pre-existing or simulated trust that
is not based on past experience. It is utilised when there is no evidence from previous
interactions by which one party can assess another party’s trustworthiness (Moorman
et al. 1992). First impressions will give rise to beliefs, which in turn can become
convictions. According to Rempel et al. (1985), people expect that future events will
prove their convictions to be correct. Faith is seen as an emotional security that allows
individuals to go beyond shared experiences and hope they will not be harmed by
entering into a new relationship. Reputation is based on information about a potential
party’s behaviour (Malaga 2001; McKnight et al. 1998; Yamagishi and Yamagishi
1994; Sztompka 1999). This kind of information is not as reassuring as direct
information, which arises via an accumulation of experience within a certain situation.
According to McKnight et al. (1998), parties with a strong and positive
reputation are perceived as being trustworthy, and parties with a poor reputation are
seen as untrustworthy. However, in order to be trusted, individuals endeavour to
establish a favourable reputation (Malaga 2001); all trust is explicitly based on
reputation (Dasgupta 1988:53). Furthermore, reputation can play a major role in
forming the first impressions that are held by people entering new relationships.
Although reputation does not necessarily imply direct interaction, it can be utilised as
information on which to base one’s judgement of trustworthiness (predictability and
dependability) (Coleman 1990; Sitkin and Roth 1993; Malaga 2001). Misztal (1996)
argues that beliefs (values and ethics), conforming to social expectations (reciprocity
of exchange) and formal control (rewards, monitoring and discipline) are the three
basic elements that provide the foundation for building a reputation.
Chapter Three – The Concept of Trust
-61-
3.5 Trust and Distrust
In spite of these different perspectives on trust, the achievement of trust in
relationships has been shown to have several benefits. For instance, it lowers
transaction costs (Williams 1988). However, there is also a negative side of trust,
known as distrust (Rotter 1967; Bromiley and Cummings 1995; Coleman 1990).
Distrust can be expressed as “a belief that a person’s values or motives will lead them
to approach all situations in an unacceptable manner” (Sitkin and Roth 1993:373).
Several researchers (for example, Lewicki et al. 1998; Lewicki and Bunker
1995; Sitkin and Roth 1993) have viewed distrust as the opposite of trust; although
the concepts are related, these researchers, (for instance, Lewicki et al. 1998:439)
agree that:
“…both trust and distrust are understood in behavioural terms, with little attention given to the confidences, intentions, and motives that promote trusting/distrusting and trustworthy/untrustworthy behaviour”.
However, Lewicki et al. (1998) also note that although trust and distrust are
separate terms, they have related dimensions. Further, there are different factors that
can affect distrust negatively or positively, and these depend on the individual’s
perceptions and experiences.
For psychologists, psychological conflicts are characterised by simultaneous
trust and distrust, a situation which is unstable and temporary in nature (Lewicki and
Bunker 1995). Sociologists recognise the importance of trust and distrust as
instruments for reducing social complexity and uncertainty (Lewis and Weigert
1985). Economists view distrust by focusing on ‘cheating’ in sense of the calculative
perception of trust, while viewing distrust in relationships as, for example, switching
to a an inferior product (Axelrod 1984).
Nevertheless, there are two main criticisms of the theoretical differences
between trust and distrust. Firstly, the differentiation between trust and distrust is
undermined by the observation that positive and negative assessments of influence
might represent responses to different measures. Secondly, it is insufficient to define a
Chapter Three – The Concept of Trust
-62-
trustor’s attitude in terms of its antecedents7
, because antecedents might indicate why
an individual trusts, but not how much they trust (Mayer et al. 1995). Because trust is
a personal assessment, the importance of any known antecedent to a trustor’s attitude
within a relationship will vary based on the perceived conditions of the situation in
which trust is bestowed (Lewicki et al. 1998). The importance of a precise antecedent
is not likely to differentiate between trust and distrust, because when considered from
the trustor’s perspective, the importance of any given antecedent is a personal matter.
Although trust and distrust are not different attitudes, they represent different
evaluations of influence; trust represents an optimistic positive attitude towards the
relationship, while distrust represents an unfavourable attitude. As such, trust is
balanced in nature, in that each trustor’s attitude includes a judgment about whether
influence is positive or negative. Given the inherent symmetry of this judgment, trust
and distrust represent opposing standpoints on the same underlying dimension of the
perceived quality of influence (Clark and Payne 1997; Gambetta 1988).
7 From the stand point of this thesis, it is not appropriate to define trust’s antecedents or corollaries individually, because any resulting definition of trust would be inconsistent and depend on the form/context of trust, the characteristics of the engaged parties and whichever factors can influence both the antecedents and the corollaries of trust.
Chapter Three – The Concept of Trust
-63-
3.6 Trustworthiness
As highlighted in the previous discussion, academics vary in their conceptions
of trust, including its definition and proper object. For instance, while some
academics see trust as a belief, others view it in more active terms. Hardin (2002,
2006), who is a political scientist, argues that trust is a belief about another party
based on accumulated evidence of that party’s trustworthiness. Hence, the distinction
between trust and trustworthiness is that trust is the trustor’s psychological state,
while trustworthiness is a characteristic of the trustee (Corritore et al. 2003).
In their work, Mayer et al. (1995) suggested that perceived trustworthiness is
the trustor’s perception of how trustworthy the trustee is, while trust is the trustor’s
willingness to engage in risky behaviour that stems from the trustor’s vulnerability to
the trustee’s behaviour. Trustworthiness is a characteristic of the trustee, and may
stem from several perceptions about the trustee held by the trustor.
Trustworthiness is described in the literature as the perceived probability that a
particular trustee will maintain one’s trust (McKnight et al. 2002). Hardin (2002:28)
defines trustworthiness by stating that “… your trustworthiness is your commitment
to fulfil another’s trust in you”. Putting Hardin’s (2002) argument into context,
trusting beliefs represent a “… sentiment or expectation about an exchange partner’s
trustworthiness” (Moorman et al. 1993:315). Mayer et al. (1995) defined
trustworthiness as an attribute of a trustee who is responsible for trust. According to
Mayer et al. (1995), the perceptions of the trustor that affect their perception of the
trustee concern the trustee’s ability, integrity and benevolence. Doney and Cannon
(1997) treat trust as a single construct dealing with trustworthiness, integrity, ability,
and benevolence.
According to Hardin (2002), trust is established in the context of ongoing
interpersonal relationships, because it is only through such relationships that
customers can accumulate sufficient evidence to form trusting intentions (i.e. to trust
or not to trust). For this reason, trust cannot be a generalised expectation of all people.
In contrast to Hardin (2002), Sztompka (1999) describes trust in more general terms.
He argues that trust has its basis in beliefs about another person; he indicates that
individuals use their beliefs to make a bet about a person’s future actions. In this way,
Chapter Three – The Concept of Trust
-64-
trust requires an active commitment from the trustor, who has to observe a certain
level of trustworthiness before making the decision to trust.
It is widely accepted that trust is the cornerstone in long-term relationships
(Spekman 1988), and, consequently, is a key determinant in successful relationship
marketing (Morgan and Hunt 1994). Furthermore, trust is a central factor in reducing
uncertainty and risk prior to engaging in relationships (McKnight and Chervany 2000)
and is the main driver for customer loyalty (Reichheld and Schefter 2000). However,
these benefits for both individuals and organisations (such as creating loyal
customers) occur only when trustworthiness is warranted (Baier 1986; Hardin 2002).
Therefore, it becomes necessary to evaluate a party’s trustworthiness before extending
one’s trust. Hardin (2002) explains that trustworthy people follow through on their
commitments to others. Govier (1994) identifies two dimensions of trustworthiness;
competence and motivation. When there is reason to believe that a person is both able
and willing to follow through on a commitment, we are likely to extend our trust. If
either competence or motivation is called into question, we may withhold trust. While
such a stance may be protective, it precludes the possibility of realising the benefits of
trust. Conversely, if trust is extended and either competence or motivation proves to
be lacking, we may be reluctant to extend our trust in future situations. Thus, trust is
only beneficial when it is accompanied by trustworthy behaviour.
Kumar et al. (1995) see trust in terms of honesty and benevolence. Building on
the work that has been conducted by Mayer et al. (1995), McKnight et al. (2002)
undertook a mapping exercise of the various studies on trustworthiness. They
concluded that there are three main dimensions of trustworthiness across the different
literature; these are:
• Benevolence; indications that an organisation has positive intentions towards
its consumers beyond merely generating profits8
• Integrity; confidence that the organisation has particular ethical, moral and
professional standards, such as honesty and fairness, to guide its interactions
with its customers.
, such as demonstrating
genuine care about their customers.
8 Mayer et al. (1995:717) used the term “egocentric profit motive” in their study.
Chapter Three – The Concept of Trust
-65-
• Ability; the consumer’s confidence that the organisation possesses sufficient
skills to deliver the required service.
These trusting beliefs are related but distinct. For example, consumers may
believe that an organisation cares about its customers and thus intends to deliver a
smooth, error-free transaction (in this case the organisation is benevolent), but they
may also believe that the organisation lacks the expertise to do so. Similarly, although
beliefs about integrity and benevolence are similar, the former focuses on meeting
objective standards of corporate citizenship, while the latter focuses on customer
benefits that go beyond normal business activity. For instance, even though
consumers believe that an organisation follows a professional code of conduct (i.e. it
has integrity), they may still question the organisation’s genuine concern for its
customers (i.e. its benevolence).
Even though ability, benevolence, and integrity are acknowledged as being
conceptually distinct (for example, see Kumar et al. 1995), they are often combined
into a global measure of trusting beliefs (see for instance, Doney and Cannon 1997).
Where a combination of these beliefs into a single variable is used as a simple
approach to studying trust, it can be difficult to identify what action should be taken in
order to build trust (Doney and Cannon 1997).
From a sociological point of view, Sztompka (1999) argues that enhancing
trustworthiness increases the level of trust. For this reason, Sztompka (1999)
identified three indicators of trustworthiness; reputation, performance and appearance.
Reputation is a record of past acts, and trust built on this conception can lead to
several advantages (such as increasing word-of-mouth effectiveness). Performance
represents actual acts, current conduct and those results that are observable.
Appearance provides an indication regarding trustworthiness. This is a sociological
level at which humans assume trustworthiness from appearance and pre-conceived
stereotypes. For example, certain ethnic groups may not be trusted because of a
historical incident9
9 For instance, the Germans aren’t trusted by some groups because of World War II.
(Sztompka 1999). These three indicators of trustworthiness are
important, but it appears that they need to act in concert to achieve profound
trustworthiness.
Chapter Three – The Concept of Trust
-66-
From a psychological perspective, Rousseau et al. (1998) argue that
trustworthiness encompasses four forms of attribute; competence; positive intentions,
ethics and predictability. The effect of each of these attributes is to build up the
trustor’s confidence that the trustee is willing and able to fulfil the trust.
• Competence: Butler (1991) argues that competence requires that the trustee
possesses the knowledge, expertise or skill to fulfil the needs of the trustor. A
related attribute is credibility, which implies the extent to which information
offered by the trustee can be believed (Doney and Cannon 1997).
• Positive intentions: according to Mayer et al. (1995), these symbolise the
trustee’s feelings toward the trustor. Positive intentions may include goodwill
(Blomqvist 1997), benevolence (McKnight et al. 1998) and customer loyalty
(Butler and Cantrell 1984).
• Ethics: the honest values to which the trustee adheres; these values differ from
positive intentions in that both parties are engaged toward others in general,
rather than toward a precise trustor. Ethical values concerning trust include
integrity (Mayer et al. 1995), honesty (Rampel and Holmes 1985), fairness
(Butler 1991), mutual commitment (Morgan and Hunt 1994) and the
achievement of any stated commitments (Zaheer et al. 1998).
• Predictability: the extent to which the trustee’s behaviour matches the other
party’s expectations (Butler 1991; Zaheer et al. 1998). This is similar to
reliability (Sheppard and Sherman 1998) or consistency (Butler 1991).
However, these expectations are often based on previous experience of the
other party’s behaviour, but both engaged parties might also derive
predictability information from expectations related to a particular social
position (Doney and Cannon 1997).
Mutual competence and predictability are naturally related to the cognitive
dimension of trust (McAllister 1995), while positive intentions are normally related to
the affective dimension of trust. In addition, beliefs concerning the ethics of a trustee
may be embedded in either the cognitive or the affective dimensions of trust,
depending on whether they are based on objective evidence or on an emotional bond
(Mayer et al. 1995).
Chapter Three – The Concept of Trust
-67-
Based on both Rousseau et al.’s (1998) and Mayer et al.’s (1995)
categorisation and description of trustworthiness, it can be argued that trustworthiness
relates to the set of ‘confident positive expectations’ customers have about the
intentions and expected future actions of an organisation that they are intending to
interact with (Lewicki et al. 1998). In other words, trustworthiness is a trait of a party
in a relationship that inspires confidence for those who rely on the good intentions of
others to perform services competently and in their best interests (Caldwell and
Clapham 2003).
In a marketing sense, trustworthiness is a key determinant for successful
relationships (Butler and Cantrell 1984; Mayer et al. 1995; Lewicki et al. 1998;
Morgan and Hunt 1994; Rousseau 2005). However, there is a lack of unified
understanding in the literature concerning the differences between trust and
trustworthiness and how both constructs should be perceived (Gefen et al. 2003). This
thesis views trustworthiness as a personally formed value judgment about the
behaviour of another party who wants to be trusted. This view is aligned with the
some of the well established theories on trustworthiness (for example, Glaeser et al.
1999; Bews and Rossouw 2002; Caldwell and Clapham 2003).
Hardin (2002:29) noted that “… much of the literature on trust hardly
mentions trustworthiness, even though implicitly much of it is primarily about
trustworthiness, not about trust”. Hardin’s (2002) argument was further supported by
Caldwell and Clapham (2003) when they argued that the decision to trust is made
based upon one party’s assessment of the behaviour of another party, through what
Caldwell and Clapham (2003) described in their proposed model as a ‘mediating lens’
or complex filter through which each person views the world. In other words, the
decision to trust has to pass through a trustworthiness filter. However, trustworthiness
becomes an even more salient assessment in high-risk transactions (Eysenbach et al.
2002). Furthermore, the majority of scales used to measure trust (for example,
Sirdeshmukh et al. 2002) are research specific (Büttner and Göritz 2008). These
scales do not distinguish whether perceived trustworthiness is a uni- or
multidimensional construct.
Chapter Three – The Concept of Trust
-68-
Despite efforts to map the determinants of trust in the literature (for example,
see Rousseau et al. 1998), there is still a lack of general agreement about these
determinants. This relates to the fact that most of these studies did not take into
consideration the contribution of relationship marketing10
This thesis follows the argument put forward by Hardin (2002) and
empirically tested by Caldwell and Clapham (2003), according to which the majority
of the existing literature on trust is actually about trustworthiness. Hence, the
dimensions of trustworthiness are related, if not identical, to the dimensions of trust
(for instance, shared values). This implies that the perceived benefits of trust can also
be seen as perceived benefits of trustworthiness; this will be discussed in the
following chapter on conceptualising trustworthiness
.
11
10 Some studies did take into consideration the contribution of relationship marketing; however, these studies are rather limited (for example, Morgan and Hunt 1994). 11 In this thesis, however, for clarity the term trust will be used hereafter to indicate trustworthiness unless otherwise stated.
.
Summary
Trust is a very complex term, with several levels and determinants. Therefore,
it is unwise to use the term ‘trust’ and assume its meaning is fully and properly
understood within one perspective, due to the term’s different meaning in another
perspective. While the concept of trust in a marketing sense attracts major attention
within the marketing literature, our understanding of the construct can be both varied
and ambiguous. Most authors in marketing research draw upon a belief-based
conceptualisation of trust – that is, perceived trustworthiness. Furthermore, there is no
published work incorporating trustworthiness as a conceptual construct (with the
exception of Gefen (2002) and Caldwell and Clapham (2003)), despite Mayer et al.’s
(1995) conceptual emphasis on its importance. The next chapter will rationalise the
foundations of the conceptual framework for the current research, which aims to
identify the antecedents and outcomes of trustworthiness.
Chapter Four – Justification of the Conceptual Model and the Research Hypotheses
-69-
CHAPTER FOUR: JUSTIFICATION OF THE
CONCEPTUAL MODEL AND THE RESEARCH
HYPOTHESES
4.1 Introduction
The concept of trust has been examined in various contexts over the years; for
instance, research has been carried out in industrial buyer-seller relationships (Doney and
Cannon 1997), distribution channels (Dwyer et al. 1997) and market research (Moorman et
al. 1993). Other researchers conceptualised trust as a foundation for interpersonal
relationships (Rempel et al. 1985; Svensson 2001) and as the basis for stability in
organisational operations (Rousseau et al. 1998). However, most academics agree that
relationships that are viewed as highly trust-oriented (or affective) are greatly valued by the
exchange parties (see for example, Swan et al. 1985; Zaltman and Moorman 1988;
McAllister 1995). It seems that the trustee is more willing to commit to a relationship if trust
is present (Morgan and Hunt 1994). Moreover, some organisations employ trust as a method
of reducing risk and uncertainty (Young 2006).
Despite the general agreement on the benefits of trust, there are several perspectives,
in the literature, regarding the nature of trust, as discussed in the previous chapter. Each
perspective characterises trust to fit within its domain (for instance, the contrasting views in
economics and psychology). Furthermore, each proposes a different cluster of characteristics
within which trust can be formed. In spite of the variations in the different academic
perspectives, however, the overall objective is to provide an appropriate understanding of the
concept of trust in order to achieve a competitive advantage. Hence, all of these schools of
thought agree about the importance of developing trust to produce efficiency, productivity
and effective relationships. As discussed previously1
1 Chapter Three - Section 3.6.
, the various strands of literature
Chapter Four – Justification of the Conceptual Model and the Research Hypotheses
-70-
acknowledged the importance of trustworthiness as a key factor for successful long term
relationships. Therefore, this thesis proposes a conceptual model of trustworthiness. The
model explains how trustworthiness is formed by proposing several antecedents to
conceptualise trustworthiness and its hypothesised outcomes.
4.2 Characteristics of Trust
Academics from a range of social science disciplines have attempted to define the
nature of trust (see Table 4.1 for a review of the key authors). For example, Butler and
Cantrell (1984) proposed five characteristics of trust, which are: integrity (the reputation for
honesty and truthfulness on the part of the trusted individual); competence (the technical
knowledge and interpersonal skills needed to perform the job); consistency (reliability,
predictability and good judgment in handling situations); loyalty (benevolence, or willingness
to protect, support and encourage others); and openness (mental accessibility, or the
willingness to share ideas and information freely with others). Butler and Cantrell (1984)
proposed that trust emerges through these characteristics, although the precise mixture of the
five characteristics in any situation might vary depending on the exchange parties’ personal
acceptance of trust.
Table 4.1: Classification of Trust by Discipline with Key Authors Meanings Marketing Lit. Management Lit. Psychology and
Sociology Lit. Expectation
Garfinkel (1967) Dwyer et al. (1987) Doney and Cannon (1997)
Zucker (1986) Bhattacharya et al. (1998) Zaheer et al. (1998)
Boon and Holmes (1991) Rotter (1971; 1980)
Belief
Anderson and Weitz (1989) Crosby et al. (1990) Anderson and Narus (1990) Piercy and Morgan (1991) Andaleeb (1995) Doney and Cannon (1997)
Robinson (1996) Lewicki et al. (1998) McKnight et al. (1998)
Mistzal (1996)
Willingness
Butler and Cantrell (1984) Andaleeb (1992) Moorman et al. (1992) Andaleeb (1996)
Zand (1972) McAllister (1995) Mayer et al. (1995) Doney and Cannon (1998)
Lewis and Weigert (1985)
Chapter Four – Justification of the Conceptual Model and the Research Hypotheses
-71-
Doney and Cannon (1997)
McKnight et al. (1998)
Confidence
Butler and Cantrell (1984) Butler (1991) Morgan and Hunt (1994) Kumar et al. (1995) Doney and Cannon (1997)
Ring and Van de Ven (1992) McAllister (1995)
Rempel (1985) Rempel and Holmes (1986)
Attitude
N/A* Whitener et al. (1998) Luhmann (1988)
Reliance
N/A Curral and Judge (1995)
Giffin (1967)
Acceptance/Incorporation of risk/vulnerability
Lewicki et al. (1998) Koller (1988)
Williamson (1993) Sheppard and Sherman (1998) Baier (1986)
Coleman (1990)
Psychological state
N/A Rousseau et al. (1998)
Gibb (1964)
Perception
N/A Anderson et al. (1987) Doney and Cannon (1997).
N/A
Subjective probability
N/A Nooteboom (1996)
Gambetta (1989)
Feeling
N/A Tyler and Degoley (1996)
N/A
Assumption
N/A Robinson (1996)
N/A
Judgement N/A Webb (1996) N/A
*N/A = Not Available
(Source: Adapted from Castaldo 2003)
Butler (1991) extended the earlier work by Butler and Cantrell (1984) on the
characteristics of trust by adapting the characteristic of general benevolence into an implicit
promise from a party to fulfil its obligations, and not to mistrust the other engaged party;
Butler (1991) also focused on customer loyalty as a main dimension of trust. Other academics
had the same idea as Butler and Cantrell (1984) and tried to identify the nature of trust (see
Table 4.1). For instance, Lewicki et al. (1998) identified three key dimensions in their model
to describe the fundamental characteristics of trust: risk, vulnerability and interdependence.
Chapter Four – Justification of the Conceptual Model and the Research Hypotheses
-72-
These characteristics can be seen as an essential key to understanding the nature of trust, and
are accepted as the main drivers for trust formation. They are defined below.2
1. Risk
Sztompka (1999) proposes that their very nature trust and risk are inextricably
associated with each other. If risk doesn’t exist, there is no need for the existence of trust
(Sztompka 1999)3. Despite the diverse nature of trust, there are two important elements that
are always associated with it; these are uncertainty and risk (Moorman et al. 1992; Mayer et
al. 1995; Rousseau et al. 1998). Koller (1988:265) proposes that the level of trust increases
“…as the degree of risk that is present in the situation becomes higher”. In addition, “…risk
creates an opportunity for trust, which leads to risk taking” (Rousseau et al. 1998:393).
However, since it is a basic logical premise that if there is no risk there is no need for trust,
the existence of trust reduces risks (Mayer et al. 1995; Rousseau et al. 1998). In other words,
trust would not be necessary if actions could be accepted with complete confidence and no
risk. Within a relationship context, the trustor and trustee interaction is characterised by
dependency4
under specific levels of uncertainty and risk. Some scholars have distinguished
between trust, defined as the willingness to undertake risk, and trusting behaviour,
demonstrated by the actual assumption of risk (Mayer et al. 1995). The decision to form trust
can be affected by cognitive and emotional elements (McAllister 1995). Using rational
thinking, an individual will form criteria for calculating the different aspects, and preferably
the positive advantages, of any relationship. Hence, a calculation of risk utility is often
employed, estimating the potential loss or profit and multiplying by the total probability of
that situation occurring. However, people do not always take the most rational decisions
(Lewicki et al. 1998).
2 Lewicki et al.’s (1997) characteristics were adapted to include more recent literature. 3 This is well illustrated by the economic view of trust, particularly free market theory. 4 Ideally this is a mutual dependency. However, in reality the dependency level varies depending according to the context of the relationship and the characteristics of the engaged parties.
Chapter Four – Justification of the Conceptual Model and the Research Hypotheses
-73-
2. Vulnerability:
“Being vulnerable implies that there is something of importance to be lost … making
oneself vulnerable is taking risk” (Mayer et al. 1995:712). Economists have viewed the
vulnerability feature of trust from three main angles: 1) market vulnerability; 2) investment
vulnerability; and 3) transactional vulnerability. Market vulnerability is similar to
environmental or primary uncertainty (Williamson 1993). Investment vulnerability has been
linked to the transaction costs of conducting an exchange; further, investment vulnerability
represents the risk of the transaction process, or what Williamson (1993) called asset
specificity; transaction validity is concerned with the interaction in the buying selling
process. In a marketing sense, vulnerability is a key dimension of trust, since individuals
must be willing to be vulnerable in order to trust (Mayer et al. 1995). Other researchers have
focused on vulnerability implicitly through the reliance on an exchange partner (Chaudhuri
and Holbrook 2001), and explicitly by stating that “…without vulnerability, trust is
unnecessary” (Moorman et al. 1992:82).
3. Interdependence:
Lewicki et al. (1998) notes that interdependent relations are complex. They argue that
when individuals are dependent on others, they have to accommodate the demands of the
other exchange parties. When the involved parties are interdependent, however, they have an
opportunity to influence each other (this can be a positive or negative influence – regarding
the decision to purchase a service or not to deal with a certain service provider, for instance).
Interdependent relationships are characterised by connected aims, so that the parties need
each other to accomplish their goals. Lewicki et al. (1998) comment that the possession of
interdependent goals does not mean that everybody shares the same objectives. The latter
may comprise different personal or group objectives. At times these goals and objectives may
even be contradictory. Nevertheless, Sheppard and Sherman (1998) argue that both risk and
interdependence are essential for producing trust, but the very nature of risk and trust may
well change as the level of interdependence in relationships increases risk decreases.
Chapter Four – Justification of the Conceptual Model and the Research Hypotheses
-74-
Even though Lewicki et al. (1998) produced their work in the late 1990s, it is evident
from their research that they were greatly influenced by Dwyer et al.’s (1987) work on
interdependent relationships when they suggested structuring the interdependence process
into five stages:
1. Full mutual awareness.
2. Exploration, a phase in which the buyer tests the seller.
3. Expansion, during which contacts, interdependence levels and opportunities for
expanding the relationship are more frequent.
4. Commitment, characterised by high mutual satisfaction, trust and (implicit and
explicit) commitment to maintain the relationship.
5. Dissolution of the relationship, determining the end of the relational experience, often
caused by unilateral behaviour.
Based on the fact that trust is a key factor in human interactions, and more precisely
in market exchange as part of the process of developing and maintaining customer
relationships in any organisation, the formation of trust relies on shaping the trustee’s
expectations about the motives and behaviours of the trustee (Doney and Cannon 1997).
However, since service organisations provide intangible services, the formation of trust might
seem to be relatively difficult, because different customers perceive services differently and
so the nature of trust will change accordingly.
Despite the fact that there are several schools of thought within the trust literature (see
Chapter Three section 3.5), the characteristics of trust mentioned in Table 4.1 generally
overlap between them. For example, psychology research can adopt the same
conceptualisation as marketing. Larzelere and Huston (1980) treat trust in terms of
benevolence and integrity, while Rempel (1985) and Doney and Cannon (1997) regard trust
as a combination of integrity, benevolence, ability, and trustworthiness. This overlap makes
trust difficult to measure from the point of view of one single perspective. Therefore, this
thesis does not adopt the view of any particular discipline regarding how to perceive trust or
trustworthiness, it, however, draws on several views, for instance, economics, sociology and
psychology. This allows the proposed model to benefit from the various strands of literature
without being biased towards a particular domain.
Chapter Four – Justification of the Conceptual Model and the Research Hypotheses
-75-
The main debate emerging from the above discussion is not about whether the
different schools of thought overlap in their views or not5; neither Mayer et al. (1995) nor
McKnight et al. (1998) are entirely accurate in their propositions6
4.3 Advantages of Trustworthiness
on how to conceptualise
trust. Instead, the issue is that there is confusion about whether the dimensions characterise
trust or trustworthiness. This is evident from various studies on trust in which researchers’
utilised dimensions which explicitly include trustworthiness in their definition of trust (for
example, see Gefen (2000), Tung et al. (2001), Grazioli and Wang (2001) and Pavlou 2001).
The specific wording of the items varies in different studies. For instance, Tung et al. (2001)
use the phrase ‘trustworthiness of vendor’, while Grazioli and Wang (2001) ask ‘whether the
subject felt that the store was trustworthy’. Gefen (2000:740) included the statement ‘I
believe that Amazon.com are trustworthy’ in their initial conceptualisation of trust, but
removed it after it did not achieve the required level of significance in their structural
equation modelling analysis. It is clear that in these studies there is no distinction between
trust and trustworthiness, and there is little justification for why these items were chosen.
Therefore, this thesis treats trust and trustworthiness as different concepts, although they are
similar in their meaning unless stated otherwise.
As discussed in Chapter Three section 3.4, the literature on trust often mixes trust
with trustworthiness, a confusion which Hardin (2002) illustrates in his book. Therefore, the
benefits of trustworthiness are seen to be similar to the benefits of trust in the various strands
of the literature. This is the argument this thesis has considered when dealing with the
benefits of trustworthiness.
Determining the advantages of trust in any context should be associated with the level
of trust, because each level of trust has associated expectations about the results of that trust.
The advantage of trust for any organisation is illustrated by Fukuyama (1995: 27), who noted
that:
5 It noted that online trust and trustworthiness is a separate field of study which is beyond the scope of this thesis. 6 See section 4.5 for more detailed discussion on Mayer et al.’s (1995) work.
Chapter Four – Justification of the Conceptual Model and the Research Hypotheses
-76-
“If people who have to work together in an enterprise trust one another because they are all operating according to a common set of ethical norms, doing business costs less. Such a society will be better able to innovate organizationally, since the high degree of trust will permit a wide variety of social relationships to emerge. Hence highly sociable Americans pioneered the development of the modern corporation in the late nineteenth and early twentieth century, just as the Japanese have explored the possibilities of network organisations in the twentieth.”
The importance of trust can be explained by the fact that “we trust [others] to use their
discretionary powers competently and non-maliciously” (Baier 1986:240), and it is seen as a
phenomenon which contributes to the strength of interpersonal relationships as well as intra-
and inter-organisational relationships (Svensson 2001:431). Trust is seen as a central factor
for all organisational transactions (Dasgupta 1988); it enhances relationships within the
organisation (Shapiro 1987), reduces operational costs (Williams 1993) and improves the
organisational decision-making process (McAllister 1995). From an economist’s point of
view, trust is often a prerequisite for:
1. Lowering transaction costs (Dyer 1997).
2. The development of superior information systems (Dyer 1997).
3. Assisting investments in business transactions.
According to Rousseau et al. (1998), the qualities of trust can be observed at different
levels within an organisation. At a macro level, trust is considered to be the basis of a stable
collective existence, “since complete contracting is impossible, and some degree of
uncertainty is unavoidable” (Ennew and Sekhon 2003:8). At a micro level, the formation of
trust can be seen as a method of for “promoting cooperation and reducing conflict in a variety
of different settings, including within work groups, between employees and managers and
across organisations” (Ennew and Sekhon 2003:8). In brief, the benefits of trust to
individuals and organisations vary according to the complex nature of trustworthiness and the
different views of how trust is formed.
Chapter Four – Justification of the Conceptual Model and the Research Hypotheses
-77-
4.4 Dimensions of Trust in The Literature
The theory surrounding the concept of trust examines several factors that are seen as
the antecedents of trust, and, hence, they affect the outcome of the trust; this is often
mentioned in terms of relationship commitment (Morgan and Hunt 1994) or customer loyalty
(Reichheld and Sasser 1990). However, the antecedents of trust may include several
dimensions depending on the nature of the research setting (see Figure 4.1).
Figure 4.1: Antecedents and Outcomes of Trust
(Source: Adapted from Castaldo 2003)
Main trust antecedents Experience
The other party’s skills and
competencies
Non-opportunistic motivations and
benevolence
Subject’s personal characteristics
Fairness and justice
Cooperative behaviour
Communication
Other antecedents
Level of specific investment
Similarities between partners
Value sharing
Organisational structure and culture of
the partners
Features of the transaction object
Brand value
Partner’s degree of customer orientation
Trust
Trust
Main trust consequences Uncertainty reduction
Commitment increase
Conflict containment
Non-coercive power
Fairness and justice
Satisfaction
Cooperative behaviour
Communication
Loyalty
Other consequences
Probability to allocate resources
Length of relationship
Easiness to convince the other party
Greater sales
Outcomes
Antecedents
Chapter Four – Justification of the Conceptual Model and the Research Hypotheses
-78-
Figure 4.1 highlights the main determinants of trust and the predicted consequences. It
can be argued that the key common determinants of trust are:
1. Past experiences with the other exchange party, i.e. the length of the relationship.
2. “The trustee’s perceived capabilities and competences, allowing him/her to act in line
with the expectations” (Castaldo 2003:193).
3. Both the trustor and trustee pursue a joint aim without opportunistic behaviour (i.e.
expressing a degree of benevolence); this can also be expressed as shared values.
4. The trustor’s perception of the trustee’s integrity.
According to the literature7
• Levels of specific investment made in the relationship (Ganesan 1994).
these are the key determinants of trustworthiness.
However, Castaldo (2003:193) pointed out that “in situations characterised by high content of
interpersonal contacts these determinants are heavily influenced by behaviours and personal
characteristics of the subjects involved in the relationship”; for instance, characteristics such
as past experience and communication. It is worth noting that the antecedents shown in
Figure 4.1 are not ranked according to importance; they influence trustworthiness depending
on the context and the involved parties. For example, communication can be more important
than integrity in situations where the trustor and the trustee speak different languages.
In addition to the antecedents mentioned in Figure 4.1, several other determinants
have been mentioned in the literature, although they are considered residual because they are
rarely quoted and closely related to the elements previously discussed. These residual
determinants are:
• Similarity between the partners (Crosby et al. 1990).
• Reputation (Ganesan 1994).
• Organisational structure and culture of the partners (Moorman et al. 1993).
• Features of the transaction objects, in particular the degree of complexity and the
difficulty of evaluating its qualitative level (Moorman et al. 1993).
• Brand value (Schurr and Ozanne 1985).
• The partners’ degree of customer orientation (Swan et al. 1985). 7 See section 4.6.1 in this chapter.
Chapter Four – Justification of the Conceptual Model and the Research Hypotheses
-79-
In light of this examination of trust dimensions in the literature, it appears that several
common themes can be used to explain the nature of trustworthiness in past research8
1. Common values; the level to which the engaged parties have beliefs in common
concerning what behaviours, attitudes, aims and procedures are significant or
insignificant (Morgan and Hunt 1994).
as
shown in Table 4.2, these themes are:
2. Communication; encourages trust by supporting the resolution of disputes and doubts,
and in aligning perceptions and expectations (Moorman et al. 1993).
3. Opportunistic behaviour; associated to the transaction cost theory9
.
8Even though these researchers did not discuss trustworthiness precisely, they often provided some indication of trustworthiness in their exploration of the concept of trust. This is explained by (Hardin 2002), as discussed in the previous chapter. 9 See Chapter Three - Section 3.3.2 for a more detailed discussion of transaction cost theory.
Chapter Four – Justification of the Conceptual Model and the Research Hypotheses
-80-
Table 4.2: Key Themes of Trust
Criteria Typologies
Trust dimensions
Ideological, cognitive, emotional and routine trust (Lewis and Wiegert 1985) Affective, cognitive and behavioural trust (Cummings and Bromiley 1996) Behavioural and intentional trust (Nooteboom et al. 1997) Affect-based and cognition-based trust (McAllister, 1995) Reliability and emotional trust (Johnson-George and Swap, 1982) Institutionalisation and habitualisation (Nooteboom et al. 1997) Competence- and goodwill-based dimensions (Nooteboom 1996)
Relational layer
Calculative, institutional (‘hyphenated’) and personal trust (Williamson 1993) Individual, inter-personal, institutional trust (Lewicki and Bunker 1995) Calculative, relational and institutional trust (Rousseau et al. 1998) Dispositional, personal/interpersonal and system trust (McKnight and Chervany 1995)
Contents and antecedents
Calculative, knowledge-based and institutional (Lewicki and Bunker 1996) Deterrence-based, knowledge-based and identification-based trust (Shapiro et al. 1992; Sheppard and Tuchinsky 1996) Predictability and explorative trust (Huemer 2000)
Development
processes
Characteristic-based, process-based, institutionally-based (Zucker 1986) Calculative processes, predictive processes, intention-based processes, knowledge-based processes, transfer-based processes (Doney and Cannon 1997; Doney et al. 1998)
Other classifications
Basic trust, guarded trust and extended trust (Brenkert 1998) Deterrence, obligation, discovery and internalization (Sheppard and Sherman 1998) Task-focused, fiduciary and relational forms of trust (Barber 1983)
Contiguous concepts
Trust, faith, confidence and reputation (Luhmann 1988; Hart 1989) Trust, power and commitment (Gambetta 1989; Anderson and Weitz 1993; Morgan and Hunt 1994) Rational prediction, probable anticipation, uncertainty, panic, fate, faith (Lewis and Weigert 1985)
(Source: Adapted from Castaldo 2003)
It can be concluded from Table 4.2 that there are wide gaps between the emerging
themes of trust in the literature, with the consequence that agreeing on one perspective for the
study of trust is difficult. However, there appears to be a general agreement regarding the
existence of cognitive and affective levels for the dimensions of trustworthiness.
Consequently, the proposed model acknowledges these levels, as will be discussed below.
Chapter Four – Justification of the Conceptual Model and the Research Hypotheses
-81-
4.5 The Production of Trust
Academics hypothesised different models to conceptualise trust in different domains.
For instance, one of the classic trust models was proposed by Mayer et al. (1995). Their
model characterised trust as a willingness to be vulnerable to another exchange party (see
Figure 4.2).
Figure 4.2: Mayer et al.’s (1995) Model of Trust
(Source: Mayer et al. 1995:715)
Mayer et al. (1995) argued, in their model (Figure 4.2), that trust is based on two
types of antecedents: (a) a tendency to trust; and (b) a cluster of three views regarding the
other person’s trustworthiness, incorporating their ability, benevolence and integrity. This
model also includes risk as a mediator of the relationship between trust and the perception of
risk. Figure 4.2 illustrates the three factors of perceived trustworthiness according to Mayer et
al. (1995), these are:
Chapter Four – Justification of the Conceptual Model and the Research Hypotheses
-82-
1. Ability refers to the capacity to perform the predicted action. Whitener et al. (1998)
express the elements of trust, including ability elements such as not avoiding the issue
and getting on with the job.
2. Benevolence is a sense by virtue, according to which one party has the other’s welfare
at heart and so will not act to damage it.
3. Integrity means that the trustee follows a set of desirable principles10
From a psychological perspective, Koller (1988) proposed three ways to measure trust
in relationships. These are:
.
• Measuring the individual’s dependence or expectation that the other party will take
actions towards this interaction (this element was later developed further by Mathe
and Shapiro (1993)).
• Assessing several dimensions displayed by individuals (for instance, reliability,
honesty and sincerity).
• The creation of a number of scenarios (notional situations) and examination of
individual reactions during simulations.
Lewicki and Bunker (1996) proposed a model of the progression of trust through
different stages. At the first stage in the model, individuals are keen to take risks when
entering into dependence relationships with others since they know of the existence of
institutional safeguards (Lewicki and Bunker 1996). If the validity of the trust is managed
and repeated interaction and transaction is encouraged, the actors will develop a knowledge
base about each other, thereby creating the conditions for a transition to cognitive trust. At
this stage the other partner in the relationship has proven to be consistent and reliable; the
trustor can feel comfortable because the partner is predictable.
The accumulated experience of a calculative relationship is crucial for individuals to
form cognitive trust. When their acquired mutual knowledge has deepened and mutual
confidence has developed, there may be a new transition to normative trust based on shared
values and identification (Lewicki and Bunker 1996). The actors may be encouraged to
10 McKnight et al. (1998) built upon this model and included a fourth dimension the predictability of the trustee’s behaviour.
Chapter Four – Justification of the Conceptual Model and the Research Hypotheses
-83-
identify with each other’s goals and interests, and mutual benefit will probably emerge. At
this stage the engaged parties may become friends, but this form of trust is not very common
in business transactions where there is usually a difference in interest. Following the
production of trust in relationships, both parties will shift the relationship to a higher level –
what scholars describe as relationship commitment or customer loyalty. However, in some
situations this shift might have a negative direction, a transition that is sometimes described
as resolving the relationship11
• The nature of the actors involved in the relationship, be they customers, suppliers or
competitors;
.
Despite the fact that some academics have treated trust as a single dimension that can
be observed directly without the need to use indicators to help measure it (for example, Rotter
1971), other researchers began to conceptualise trust as a multidimensional construct (for
instance, Rousseau et al. 1998; White 1992; Williamson 1993). Parvatiyar and Sheth
(1997:241) examined some criteria of trust in their work, based on:
• The relational objective, which can be configured either as strategic or
tactical/operational; and,
• The orientation of the firm, which is defined as either relational or transactional.
Sheppard and Sherman (1998) adopted a similar approach, but in a different context;
they aimed to explore the different trust requirements in various relational settings. They
differentiated between inter-organisational relations on the basis of more abstract dimensions
than those examined by Parvatiyar and Sheth (1997), considering different relational forms
and the levels of interdependence that are established between parties. In their discussion of
these two dimensions, Sheppard and Sherman (1998) proposed, four different typologies
were defined and reported (shown in Figure 4.3).
11 For more details, see Chapter Three - Section 3.5.
Chapter Four – Justification of the Conceptual Model and the Research Hypotheses
-84-
Sheppard and Sherman (1998) suggest a grammar of trust as a foundation to allow a
better understanding of the extent to which trust can be formed and adopted. The grammar
involves relational form, depth and risks. Relationships vary in terms of their form and depth:
“… we can conceptualise relational form as either dependent or interdependent and relational
depth as either shallow or deep” (Sheppard and Sherman, 1998:424). Previously, Fiske12
12 In Sheppard and Sherman (1998:423).
(1990) had defined four essential forms of human relationships: (1) communal sharing, (2)
authority ranking, (3) equality matching, and (4) market pricing. These forms establish an
initial foundation for understanding the nature of human relationships, thereby producing
trust. In their projected model, Sheppard and Sherman (1998) built on the work by Fiske
(1990) and proposed four different forms of trust as shown in Figure 4.4.
(Source: Sheppard and Sherman 1998:426)
Figure 4.3: Various Relationship Forms
Chapter Four – Justification of the Conceptual Model and the Research Hypotheses
-85-
Figure 4.4 Forms of Dependence, Risk and Qualities of Trustworthiness
As can be noted from Figure 4.4 Sheppard and Sherman (1998) associated qualities of
trustworthiness with four types of relationship dependence (shallow dependence, deep
dependence, shallow interdependence and deep interdependence), they also conceptualised
different types of risk as a moderating factors for these relationships. These forms of
dependence are:
Shallow dependence: while forming trust in any relationship, Sheppard and Sherman,
(1998:424) stated that “… it is most meaningful to approach interdependence from the
vantage point of the person or entity engaged in trusting behaviour (i.e. in agency theory the
‘principal’, and in this article the ‘trustor’).”
Shallow interdependence: two perspectives are associated with shallow
interdependence. Firstly, when one individual is attempting to manage the relationship,
behaviour seems to be interdependent (Sheppard and Sherman, 1998). Secondly, when an
individual engages in trusting behaviour, there seems to be unidirectional dependency. Still,
Sheppard and Sherman, (1998) suggest that the two main risks for the trustor in simple
dependence are unreliability and indiscretion.
Deep dependence: “… in deep dependence relationships a trustee’s behaviour is often
outside the trustor’s purview and, therefore, is difficult to monitor” (Sheppard and Sherman
Chapter Four – Justification of the Conceptual Model and the Research Hypotheses
-86-
1998:424). Within this form of relationship there is a great focus in the transaction cost
economics literature on ‘cheating’ from a customer point of view. In other words, Sheppard
and Sherman (1998) applied economic theory to this form of relationship.
Deep interdependence; “… in deep interdependence relationships the capacity of
parties to communicate is essential” (Sheppard and Sherman, 1998:424). However, under
certain circumstances communication between the engaged parties is not always possible (for
instance, when there is a large distance between the parties). The key risk in deep
interdependence relationships is failure of anticipation; that is, the risk that one party will not
be able to fulfil the other party’s needs or actions or that one party will not be able to produce
commitment towards the relationship (Sheppard and Sherman 1998).
In an earlier work, Andaleeb (1992) proposed four typologies of trust based on
crossing the dimensions later used by Sheppard and Sherman (1998). These typologies
comprise a ‘trust continuum’, between the two extremes of strong trust and distrust (see
Figure 4.5).
Discussing this ‘trust continuum’, Andaleeb (1992) proposes, four trust typologies
classified on a trust continuum (Figure 4.5), between the two ends of strong trust and distrust.
The first is full trust, where the trustee has no opportunistic motivations and possesses a set of
competencies that are coherent with the task. Distrust is characterised by the low presence of
both these dimensions. The intermediate cases are represented by two typologies defined as
hopeful trust and unstable trust. In the first situation the trustor is convinced of the other
Figure 4.5: Trust Continuum
(Source: Andaleeb 1992:15)
Chapter Four – Justification of the Conceptual Model and the Research Hypotheses
-87-
party’s supportive motivation, but aware that the trustee does not possess the knowledge
required to properly execute the assigned task. However, the trustor hopes that the trustee will
be able to increase their knowledge base and produce the expected output. On the other hand,
in the case of unstable trust, the other party is perceived to be capable of producing the
expected results, but is moved by opportunistic motivations rather than supportive ones (see
Figure 4.6).
Recently, several frameworks have started appearing in the various strands of
literature to measure trustworthiness in particular. For example, there is the ‘trust and the
mediating lens’ framework put forward by Caldwell and Clapham (2003) (see Figure 4.7). In
their model, Caldwell and Clapham (2003) argue that the decision to trust is evaluated based
upon one party’s assessment of the behaviours of another party, through what is described as
a ‘mediating lens’ or complex filter through which each person views the world. In other
words, the decision to trust has to pass through a trustworthiness filter; one party has to
appear trustworthy in order for people to trust them.
(Source: Andaleeb 1992:12)
Figure 4.6: Classification of Different Trust Typologies
Chapter Four – Justification of the Conceptual Model and the Research Hypotheses
-88-
Figure 4.7: Trust and the Mediating Lens
This mediating lens consists of seven characteristics, as explained by Caldwell et al.
(2008:108):
• It is based upon a six beliefs model; six beliefs form the foundation for an individual’s
value system and interpretation of reality, including beliefs about self, others, the
nature of the Divine, the past, current reality, and the future (see also Caldwell et al.
2002).
• It is both cognitive and affective; perceptions include both cognitive and affective
elements that are closely related (Mackie and Hamilton 1993).
• It is contextually interdependent; the decision to trust is seen as contextually based
and profoundly complex (Mayer et al. 1995).
• It is ethically founded; trust within an organisational context involves a continuous set
of ethically-based social and psychological contracts. The subtle nature of the
assumptions that make up those contracts is often tacit and implicit (Rousseau 1995).
• It is goal directed; the purposes of perception are both instrumental and normative as
people seek to achieve goals individually, with others, and in organisations.
• It is behaviourally attributive; the common phenomenon of assigning attributive
causation to others’ behaviours applies to the lens. When behaviours are seen as
negative, a common tendency is for the observer to infer malevolent intentions or
motivations (Rousseau 2005).
(Source: Caldwell et al. 2008:108)
Chapter Four – Justification of the Conceptual Model and the Research Hypotheses
-89-
• It is systemically dynamic; the iterative process of information processing is
acknowledged to be a complete and dynamic system with inputs, outputs, analysis,
assessment, and feedback (Carver and Scheier 1998). The mediating lens is
systemically dynamic, and feedback impacts beliefs, perceptions, relationships,
attitudes, and behaviours on a continuous basis.
Caldwell and Clapham’s (2003) model did not explore trustworthiness and discuss its
dimensions; rather, they focused on the ‘mediating lens’ as the main point in their model. In
addition, they did not extend the model towards the outcome of achieving trust; their
argument is simply built on Mayer et al.’s (1995) work.
Perhaps, one of the most comprehensive marketing models in the recent literature that
explores trustworthiness as a multidimensional construct is Sirdeshmukh et al.’s (2002)
model. Sirdeshmukh et al. (2002) proposed a framework that distinguishes between
trustworthiness and trust. They used two dimensions to measure trustworthiness, described as
the cognitive and affective dimensions. Moreover, Sirdeshmukh et al. (2002) focused on
specific behavioural dimensions for two key facets – frontline employee (FLE) behaviours,
and management policies and practices (MPPs). The dimensions in Sirdeshmukh et al.’s
(2002) model are well documented within a strong theoretical framework. However, two
issues can be raised regarding the model; firstly, their model was constructed from the
organisation’s point of view in the analysis of the source of trust. This does not follow from
the principles of long-term relationship marketing, in which both the customer and the
organisation contribute equally to the relationship (Grönroos 1994). Secondly, Sirdeshmukh
et al. (2002) did not distinguish between different forms of loyalty (Rundle-Thiele and
Bennett 2001).
Based on the above, this thesis builds on the previous research on trust and
trustworthiness, and proposes a conceptual framework to measure trustworthiness. This will
be discussed in the following section.
Chapter Four – Justification of the Conceptual Model and the Research Hypotheses
-90-
4.6 Modelling Trustworthiness
As discussed previously, several researchers have attempted to explain trust and
trustworthiness as two similar concepts (Hardin 2002). However, this thesis argues that trust
and trustworthiness are different because:
• Trust represents cognitive or emotional investment in the trustee, while
trustworthiness often goes against the trustee’s self interest by putting the trustor self
interest first.
• Trust can be perceived as a risky decision, and consequently it may be dependent
mainly on risk aversion. Trustworthiness is affected by psychological, moral or
ethical factors.
• Trust is linked to the trustor’s perception of their own vulnerability to the actions of
others, while trustworthiness is not.
A number of models have been developed to describe perceived trustworthiness based
on the characteristics of a trustee that are perceived as influencing their trustworthiness
(Caldwell et al. 2008). Examples of such characteristics are the trustee’s perceived
truthfulness, ability and skill. Table 4.3 comprises several characteristics that are generally
considered to be characteristics of trustworthiness, sometimes called the antecedents of
trustworthiness (Mayer et al. 1995).
As a result of the review of different models and theoretical concepts of trust in the
literature in section 4.5 above, the main dimensions of trustworthiness – a mapping for the
dimensions of trust – has been generated, in an effort to provide a better understanding of the
importance of each construct. This is an attempt to clarify the problem identified by Mayer et
al. (1995:711) when they noted that “…one of the difficulties that has hindered previous
research on trust has been a lack of clear differentiation among factors that contribute to trust,
trust itself, and outcome of trust”. Table 4.3 presents a mapping exercise of the different trust
dimensions. These emerging dimensions will act as a working framework for the proposed
model in this thesis.
Chapter Four – Justification of the Conceptual Model and the Research Hypotheses
-91-
Table 4.3: Mapping the Dimensions of Trustworthiness
Dimensions Definition Author Communication
The formal and informal
sharing of meaningful and timely information between
two exchange parties
Dwyer et al. (1987) Anderson and Narus (1990)
Loyalty13
A partner is not only reliable,
but also performs well in extraordinary situations
(behavioural and attitudinal)
Butler and Cantrell (1984) Friedland (1990) Dick and Basu (1994) Rempel and Holmes (1995) Rundle-Thiele and Bennett (2001)
Reliability Consistency
Time and experience are critical elements in evaluating
trust
Rotter (1980) Butler and Cantrell (1984) Morgan and Hunt (1994) Curral and Judge (1995) Rossiter and Pearch (1996) Doney and Cannon (1997) Sheppard and Sherman (1998) Zaheer et al. (1998) Ennew and Sekhon, (2003)
Competence
Experience and wisdom displayed by the other partner
Butler and Cantrell (1984) Luhmann (1988) Butler (1991) Mayer et al. (1995) Ghoshal and Bartlett (1994) Cummings and Bromiley (1996) Sheppard and Sherman (1998) McKnight et al. (1998) Jarvenpaa et al. (2000) Sirdeshmukh et al. (2002)
Benevolence
Accepted duty to protect the rights of the other partner
Confidence in sharing
information or problems with the other partner
Rempel and Holmes (1981) Barber (1983) Butler and Cantrell (1984) Hart et al. (1989) Butler (1991) Ring and Van de Ven (1994) Mayer et al. (1995) Hosmer (1995) Doney and Cannon (1997) Sheppard and Sherman (1998) McKnight (1998) Gefen et al. (2003)
Multiple forms of trust Trustworthiness
There is more than one type of trust
Rempel and Holmes (1981) Butler and Cantrell (1984) McAllister (1993) Morgan and Hunt (1994) Ghoshal and Bartlett (1994) Bhattacherjee (2002) Sirdeshmukh et al. (2002)
13 Dick and Basu (1994) provided four categories of loyalty highlighted in appendix Nine.
Chapter Four – Justification of the Conceptual Model and the Research Hypotheses
-92-
Integrity
Morgan and Hunt (1994) Mayer et al. (1995) Cummings and Bromiley (1996) Sheppard and Sherman (1998) Gefen et al. (2003)
Value alignment
Sitkin and Roth (1993) Morgan and Hunt (1994) Hess (1995) Siegrist (2000)
(Source: adapted from Handfield and Bechtel 2004)
4.6.1 The Proposed Model and Research Hypotheses
This section covers the proposed model and the associated hypotheses for each of the
conceptualised construct. Primarily, the proposed model for this thesis (illustrated in Figure
4.8) assumes three main relationships, and clearly defines the determinants of trustworthiness
as well as an outcome. Secondly, the framework is a static dyadic model that assesses
relationships at a given moment. The main relationships in the model place trustworthiness at
the focal point, with determinants of trustworthiness (classified as operating at low and high
levels) having a direct relationship with the central concept. The importance of each
determinant will be discussed at a later stage in this thesis, since the results from the
empirical study will show which determinant is most significant and which has stronger
regression coefficient paths with trustworthiness14
14 Trustworthiness determinants are listed with no particular ranking.
. The third part of the model covers the
outcome of trustworthiness, which is illustrated by two types of customer loyalty, behavioural
(low level) and attitudinal (high level) loyalty, with a prediction that behavioural loyalty can
lead to attitudinal loyalty.
Chapter Four – Justification of the Conceptual Model and the Research Hypotheses
-93-
4.6.2 Trustworthiness
The model’s central focus, trustworthiness15
Butler (1991) claims that competence entails that the trustee has the knowledge,
expertise and skill to fulfil the needs of the trustor. A related attribute is credibility, which
describes the extent to which information offered by the trustee can be believed (Doney and
Cannon 1997). Mayer et al. (1995) took a different view from Butler (1991) and suggest that
positive intentions symbolise the trustee’s feelings toward the trustor. Positive intentions may
, is defined in the literature as the
perceived probability that a particular trustee will maintain one’s trust. For example, Mayer et
al. (1995) illustrate trustworthiness as an attribute of a trustee who is responsible for trust.
Trustworthiness includes four forms of attributes, comprising competence, positive
intentions, ethics, and predictability. The effect of each of these attributes is to build up the
trustor’s confidence that the trustee is willing and able to fulfil that trust.
15 In the proposed model trustworthiness has been treated as a multidimensional construct.
Benevolence
Competence
Integrity
Consistency
Value alignment
Communication
Behavioural loyalty
Attitudinal loyalty
Trustworthiness
Low level
High level
Source: Author
(Source: Author)
Figure 4.8: Conceptual Framework
Chapter Four – Justification of the Conceptual Model and the Research Hypotheses
-94-
also include goodwill (Rousseau et al. 1998), benevolence (Doney and Cannon 1997) and
customer loyalty (Butler 1991). Another element within trustworthiness is ethics, often
conceptualised as the values to which the trustee remains loyal. These values vary from
constructive intentions in that both parties are engaged toward others in general, rather than
toward a precise trustor. Furthermore, ethical values relating to trust embrace integrity
(Mayer et al. 1995), honesty (Rempel et al. 1985) and fairness (Butler 1991).
The final attribute of trustworthiness is predictability, which is the extent to which the
trustee’s behaviour meets with the trustor’s expectations in a particular interaction (Butler
1991). This is similar to reliability (Sheppard and Sherman 1998) and consistency (Morgan
and Hunt 1994). Additionally, predictability and mutual competence are naturally related to
the cognitive dimension of trust (McAllister 1995)16
4.6.3 Determinants of Trustworthiness Hypotheses
, while positive intentions are normally
related to the affective dimension of trust.
The proposed model distinguishes between two levels in the definition of
determinants of trustworthiness. High level determinants (integrity, benevolence, value
alignment and communication) deal with the emotional aspects of trustworthiness, while the
low level determinants (consistency and competence) cover the economic view of
trustworthiness, i.e. through calculating the costs and benefit of the relationship. These
determinants will be examined below.
4.6.3.1 Low Level Determinants
Determinant 1. Competence
Butler (1991) viewed competence from the stance that the trustee possesses the
knowledge, expertise or skills required to fulfil the needs of the trustor. Competence is based
on the belief that others are dependable and reliable. When individuals trust they believe that
the other exchange party will behave as expected and meet their obligations. In other words,
competence is linked to the service provider’s ability to deliver on their promises (Doney and
Cannon 1997). In contrast, customers have to be willing to accept the risks associated with
16 See the argument surrounding cognitive and affective trust in Chapter Three - Section 3.5.
Chapter Four – Justification of the Conceptual Model and the Research Hypotheses
-95-
service provider competence; in other words, customers have to show willingness to accept
that the service provider is capable of delivering what has been promised (Doney and Canon
1997; McAllister 1995).
Hypothesis 1: Competence has a positive impact on trustworthiness.
Determinant 2. Consistency
Ennew and Sekhon (2003:8) propose that “consistent behaviour over time contributes
to trust because it increases the predictability of future behaviour”. Therefore, consistency is
crucial since honesty as a single determinant of trustworthiness is unlikely to be enough to
reduce uncertainty and risk, making the achievement of trustworthiness less likely. This is
aligned to notions of honesty and promise (Butler and Cantrell 1984) to the extent that the
trustee can be confident that a trustor is honest, and will fulfil their committed obligations
(Hess 1995).
Hypothesis 2: Consistency has a positive impact on trustworthiness.
4.6.3.2 High Level Determinants
Determinant 3. Benevolence
The third determinant, benevolence, is related to the trustee’s perceived willingness to
establish mutually fulfilling interactions rather than maximising profits. Benevolence is only
present when the trustee has opportunities for opportunistic behaviour (for example,
purchasing a reduced priced service) (Mayer et al. 1995). The importance of benevolence is
illustrated by the definition put forward by Doney and Cannon (1997:36) when they defined
trust as “perceived credibility and benevolence of a target of trust”. Furthermore, affective
trust is often preceded by benevolence (Sheppard and Sherman 1998). Benevolence promotes
the effective aspects of the relationship by indicating the service organisation’s intention to
care for the well being of its customers (Doney and Cannon 1997; Mayer et al. 1995).
Therefore, it is categorised as a high level determinant in the current model.
Hypothesis 3: Benevolence has a positive impact on trustworthiness.
Chapter Four – Justification of the Conceptual Model and the Research Hypotheses
-96-
Determinant 4. Communication
Marketing academics have stressed the importance of communication between
customers and organisations. For instance, Anderson and Narus (1990) noted that the practice
of relationship marketing is centred on nothing more than good communication which leads
to trust. Grönroos (2000) recommended that organisations should create a ‘relationship
dialogue’ through participation from the engaged parties, in order to deliver the planned
message to customers – or, in other words, organisations should be able to communicate with
their customers in order to enable a high level of interaction between both parties.
Anderson and Narus (1990) argue that communication is an antecedent to trust,
indicating that it has a positive impact on trust as it assists in the resolution of conflicts and
ambiguities, as well as managing perceptions and expectations (see also Morgan and Hunter
1994; Dwyer et al. 1987). The underlying principle is that communication provides insight
into the specific roles (for example, communication initiator and receiver) that the trustor and
the trustee are willing and able to take throughout the potential interaction (Anderson and
Narus 1990). This is in line with research on trust in ongoing relationships, where it was
found that communication manages the expectations that parties have concerning the other’s
role or, for example, to keep customers updated with the organisation’s latest services
(Morgan and Hunt 1994).
Effective communication has been recognised as an important part of relationship
marketing (Dwyer et al. 1987), although these authors stated that trust is an antecedent of
communication. The current model not only embraces the standpoints of Anderson and Narus
(1990) and Morgan and Hunt (1994), who contend that open communications lead to trust,
but also argues that effective communication is a high level determinant of trust rather than
low level one.
From the organisation’s view it should be able to accept feedback from customers and
listen to their suggestions or complaints (via personal interaction) in addition to establishing a
good contact channel to deliver and market any future products or even deliver messages to
its customers. From the customers’ perspective, such a dialogue will provide them with an
effective method to express their opinions regarding the service they receive. However, the
Chapter Four – Justification of the Conceptual Model and the Research Hypotheses
-97-
dialogue should also include the possibility of learning in order to develop and maintain
communication, thereby developing the relationship between the involved parties (Grönroos
2000).
Hypothesis 4: Communication has a positive impact on trustworthiness.
Determinant 5. Value alignment
In their 1993 study Sitkin and Roth indicated that value alignment is a key factor in
developing trust. Their study associated value alignment with distrust from the perspective
that a lack of congruence in values could lead to mistrust (Siegrist 2000). Morgan and Hunt
(1994) viewed value alignment (or as they refer to it as shared values) from the perspective
that exchange parties with common beliefs about what behaviour, goals, and policies are
important or unimportant, appropriate or inappropriate, and right or wrong.
Morgan and Hunt (1994) found that shared ethical values were significant for
enhancing trust within a particular relationship exchange. Therefore, the current model
suggests that the formation of value alignment has to be initiated by the trustor, aligning
towards the trustee (Siegrist 2000).
Hypothesis 5: Value alignment has a positive impact on trustworthiness.
Determinant 6. Integrity
Integrity often implies that the trustee will demonstrate and stick to a set of principles
and values that the trustor finds acceptable (Mayer and Davis 1999). Integrity can be
expressed in terms of several concepts such as honesty, predictability, nature, credibility, and
dependability (Mayer and Davis 1999). Therefore, it is often thought of as a boundless
dimension. Trustee integrity requires that service providers extend a consistent approach to
their services. From an organisational standpoint, integrity entails that the different
departments of the organisation display a harmonised approach when dealing with the
trustor’s perception.
Hypothesis 6: Integrity has a positive impact on trustworthiness.
Chapter Four – Justification of the Conceptual Model and the Research Hypotheses
-98-
4.6.4 Customer Loyalty (Model Outcome)
In marketing theory, loyalty has often been discussed interchangeably with its
operational definition to refer to repeat purchases, preference, commitment, retention and
allegiance. Liljander and Strandvik (1996) argue that customer satisfaction and repurchase
intentions correlate positively, and it might be more beneficial to concentrate on influencing
customers’ experiences rather than altering their expectations. However, Buttle (1996)
viewed customer loyalty as an interactive feature of relationship marketing, which results in a
commitment from the customer to the organisation. Reichheld and Sasser (1990) and
Reichheld (1996) studied the value of building customer loyalty, and proposed that customers
remain loyal because of the value they receive from the service provider. Zeithaml and Bitner
(1996) argue that service quality and customer satisfaction positively affect customer
behaviour, and customers who have no service problems show the highest levels of loyalty
intentions. However, their loyalty intentions are not significantly higher than those of
customers who have experienced service problems that have been solved satisfactorily.
Therefore, organisations that are willing to improve services, particularly beyond their own
desired service level, should do so in a cost-effective manner. Organisations aiming for
customer loyalty should tailor their offers to fit customer demands.
Customer loyalty can be seen as a means to express feelings or attitudes toward
brands, services, store products and activities (Uncles et al. 2003). From the current literature
(see for example, Oliver 1999), customer loyalty can be divided into two main types, known
as behavioural loyalty and attitudinal loyalty. Attitudinal loyalty is a higher-order or longer-
term commitment from a customer towards an organisation. Behavioural loyalty, on the other
hand, addresses the customer’s intention to make repeat purchases (Mayer and Davis 1999;
Chaudhuri and Holbrook 2001; Moore and Sekhon 2005). Thus, customer loyalty in the
proposed model is seen as bi-dimensional, including both attitudinal and behavioural aspects
of loyalty. Both attitudinal and behavioural loyalty are outcomes of trustworthiness, with the
additional factor that behavioural loyalty can lead to attitudinal loyalty.
Chapter Four – Justification of the Conceptual Model and the Research Hypotheses
-99-
Outcome 1. Attitudinal loyalty
Attitudinal loyalty in this model is seen as the higher level of loyalty, and is
considered to be an antecedent of relationship commitment (Morgan and Hunt 1994).
Attitudinal loyalty is often defined as having a positive effect since it enhances a
relationship’s continuance and the need to remain in the relationship. Attitudinal loyalty is
significant because it indicates a tendency to exhibit certain behaviours, such as the
possibility of future purchase (Dick and Basu 1994). Rundle-Thiele and Bennett (2001) argue
that this type of loyalty reduces the level of expectations set by customers and the tendency to
search for alternatives (Rundle-Thiele and Bennett 2001).
Hypothesis 7: Trustworthiness has a positive impact on attitudinal loyalty.
Outcome 2. Behavioural loyalty
The repeat purchase of services or products is often seen as one of the characteristics
of behavioural loyalty, which is seen as a lower form of loyalty. It is normally accepted that
behavioural acts are an end variable in consumer behaviour models (Rundle-Thiele and
Bennett 2001). Reichheld (1996) argues that the effectiveness of relationship marketing
efforts should be evaluated in terms of the behavioural changes they bring about. As a result,
it is not surprising that behavioural loyalty is generally accepted as a relationship outcome
that will lead ultimately to attitudinal loyalty (Pine et al. 1995).
The proposed model hypothesises that trustworthiness is linked directly to
behavioural and attitudinal loyalty, with each type of loyalty as an indicator of the extent of
trustworthiness. Moreover, it is predicted that behavioural loyalty can lead to attitudinal
loyalty.
Hypothesis 8: Trustworthiness has a positive impact on behavioural loyalty.
Hypothesis 9: Behavioural loyalty has a positive impact on attitudinal loyalty (a prediction).
Chapter Four – Justification of the Conceptual Model and the Research Hypotheses
-100-
Summary
After examining several dimensions of trust, this chapter proposes a conceptual
framework to identify the components of trustworthiness within service organisations. The
model attempts to identify the antecedents of trust as a method of measuring trustworthiness,
and thereby predict some corollaries of the concept, mainly connected with customer loyalty.
Furthermore, customer loyalty is classified into two main forms, namely behavioural and
attitudinal loyalty. Each form of loyalty is linked with special attributes of the customer.
The next chapter will explore the methodological aspects of the research in order to
provide a solid base for an empirical examination of the proposed model.
Chapter Five – Research Methodology
-101-
CHAPTER FIVE: RESEARCH
METHODOLOGY
5.1 Introduction
A piece of research can be defined as any attempt to study a problem logically,
or to add to knowledge surrounding a problem; research is often implicitly assumed to
be the work of professional scientists and academics (Reber 1995). Neuman (1997)
postulates that in order for the researcher to make his or her research clear and
accessible, he or she should construct the research in such a way that it could be
carried out by non-academics, creating knowledge in the process. Research, therefore,
should become an activity that is accessible to all, instead of an activity that is carried
out by a select few people within the academic world (Neuman 1997).
All research has its own unique characteristics and ways for determining the
appropriate process to identify what the research is aiming to investigate and achieve,
i.e. how the research will contribute to the body of knowledge. The general structure
underlying any research is known as its methodology, and the process of conducting
the research is referred to as its research method.
This chapter aims to:
• Rationalise the research design of and the philosophical stand point the thesis.
• Examine the employed sampling techniques, sampling procedure and the used
measurement methods.
• Introduce the analyses strategy and the main analysis techniques for the main
data collection stage. Furthermore, the chapter will provide a full explanation
of the different research methods and strategies that will be utilised.
• Discuss the limitations of the research. (See Figure 5.1 for an overview of the
structure of this chapter).
Chapter Five – Research Methodology
-102-
(Source: author)
Several methods was used to achieve the aim and objectives of this thesis,
these are:
• Research philosophy: post-positivism
• Research logic: deductive
• Research approach: mixed methods by combining semi-structured interviews
and survey questionnaire
• Analysis method: multivariate analysis using structural equation modelling
These methods are discussed in greater detail in this chapter, each of which
has a full justification of why it was selected.
Figure 5.1: Overview of the Methodology Chapter
Research design and philosophical stand
Data collection strategy (sampling, procedure and measurement methods)
Analyses strategy
Limitations
Chapter Five – Research Methodology
-103-
5.2 The Research Philosophy
5.2.1 Choice of Philosophy
There are several competing views or paradigms concerning knowledge,
regarding how it is perceived and how it developed (Grix 2004). However, in the field
of marketing, it appears that two main paradigms dominate the majority of the
academic research; these are the positivist1
The classic distinction between positivism and interpretivism creates
corresponding distinctions in terms of ontology (the ‘reality’ being investigated),
epistemology (the relationship between reality and the researcher) and methodology
(the techniques used to investigate reality); that is to say, researchers construct
epistemological assumptions using certain methodologies to support their ontological
theories. Table 5.1 provides additional information about each of these differences.
Positivism is the dominant paradigm within the natural sciences, while interpretivism
is the dominant paradigm in the social sciences.
and the interpretivist paradigms. The
selection of a research philosophy is influenced by the researcher’s preferred method
of tackling the research (Baker 2001). The positivist paradigm views reality as
objective, existing independently of human beings (Grix 2004). The interpretive
paradigm views reality as ‘constructed’ by human beings, and consequently it does
not and cannot exist independently from the researcher who is aiming to comprehend
it. Figure 5.2 provides a general view of the process of selecting the research
methodology followed in this thesis.
1 Different scholars use different terms when referring to phenomenology and positivism, depending on the scholars’ perspectives (see Table 5.2 for more expressions).
Figure 5.2: Blocks of Research
(Source: Grix 2004:66)
Chapter Five – Research Methodology
-104-
Table 5.1: Marketing Research: Main Scientific Paradigms and Their Elements
(Source: Grix 2004)
Guba and Lincoln (1994:108) provided a simplified way of understanding the
relationships between epistemology and ontology; they argued that the researcher
should answer three questions in order to identify his or her inquiry paradigm:
The ontological question: what is the form and nature of reality and, therefore, what is
there that can be known about it?
The epistemological question: what is the nature of the relationship between the
knower (or would-be knower) and what can be known?
The methodological question: how can the inquirer go about finding out whatever he
or she believes can be known?
Some scholars (for instance, Polkinghorne 1983; Baker 2002) claim that the
choice of paradigm has a direct link with the research question(s) that has been
proposed at the start of the research, and all other methodological choices depend on
how the researcher is intending to address these question(s). Foxall (1998), using
radical behaviourism as an example, suggested that broad theories of consumer
behaviour rest on both positivist and interpretive pillars. In the case of radical
behaviourism, positivists support the search for ‘laws’ of behaviour within the
confines of laboratory settings, while interpretivists support the elaboration and
explanation of more complex consumer behaviours in natural settings. These two
dominant philosophical paradigms are discussed in detail below.
Chapter Five – Research Methodology
-105-
5.2.1.1 The Epistemological and Ontological Paradigms
The term epistemology is expressed in the literature as a “theory of
knowledge” (Harding 1987:3), or in more detail as “a theory of knowledge embedded
in a theoretical perspective” (Creswell 2003:4). Rather than asking who can be a
knower and what can be known, epistemology illustrates how knowledge is formed; it
is the foundation for the knowledge building process. The conscious and unconscious
questions, assumptions and beliefs that the researcher brings to the research, all form
the preliminary basis for an epistemological position (Creswell 2003). Actions within
the research process are influenced by the researcher’s previous experiences.
Science from the positivist tradition holds several basic beliefs about the
nature of knowledge, which together form a positivist epistemology that is the
foundation of the quantitative paradigm. Positivism entails that there is a foreseeable
reality that exists independently of the research process (Grix 2004). The social world,
like the natural world, is governed by rules, which result in patterns. Accordingly,
causal relationships between variables exist and can be identified, proven, and
explained. Thus, patterned social reality is predictable and can potentially be
controlled. This describes the nature of social reality from the positivist perspective
(Hollis 1996; Creswell 2003). In fact, both the qualitative and quantitative approaches
are located within this epistemological structure (Hesse-Biber and Leavy 2004:2), as
will be discussed later in this chapter.
5.2.1.2 The Post-positivist Paradigm
Although this thesis takes the stand point of post-positivist paradigm (as will
be discussed further on), it acknowledges the debate surrounding the meaning of
science in the literature, particularly in the social sciences, (Creswell 1994). This
debate has argued that the different methodologies that are employed in seeking
knowledge often rely on differing philosophical assumptions and principles (Neuman
1997). Therefore, each methodology employs different methods of investigation and
embodies different basic assumptions about what comprises truth and knowledge.
Figure 5.3 illustrates the position of the post-positivist paradigm in comparison to
positivist and interpretivist paradigms.
Chapter Five – Research Methodology
-106-
These paradigms place the social sciences in a complex position when it
comes to investigation. Polkinghorne (1983) illustrates this complication by reference
to the shift in thinking that has taken place over thirty years, moving research from a
predominately positivist paradigm to one that can be considered to be post-positivist
in nature. However, the positivist approach is defined here as “one which embraces
any approach which applies scientific methods to human affairs, conceived as
belonging to a natural order open to objective enquiry” (Hollis 1996:24).
The post-positivist philosophy is similar to positivism. The fundamental
difference is that when examining social reality, post-positivism recognises that
researchers cannot be completely positive about their knowledge claims (Creswell
2003:7). Building on the positivist notion of proving fundamental relationships that
compose the social world, the post-positivist approach proposes evidence to support a
pre-existing theory; in other words, post-positivists rely on deductive logic and
hypothesis testing, to create evidence that will confirm or disprove a theory, though
not in absolute terms (unlike in the positivist tradition).
Overall, post-positivist philosophy assumes that there is an objective reality
‘out there’, composed of testable cause and effect relationships (Creswell 2003).
Social reality thus exists independently of the researcher and the research question.
Relying on deductive logic, these researchers engage in measurement and hypothesis
testing in order to create evidence in support of, or against, an existing theory
(Creswell 2003).
(Source: Grix 2004:78)
Figure 5.3: The Key Research Paradigms
Chapter Five – Research Methodology
-107-
Within a post-positivist perspective, the researcher can use whatever means
are appropriate for the specific question and subject matter to be explored (Sarantakos
1998). This is possible because post-positivism provides an alternative to the
traditions and foundations of positivism in conducting a disciplined inquiry. For the
post-positivist researcher, reality is not a firm entity; instead it is a creation deriving
from those individuals involved in the research. Reality does not exist within
nothingness; its composition is influenced by its context, and many constructions of
reality are therefore possible (Crossan 2002:48).
To that extent as it has developed a conceptual model which it wishes to test;
this research is anchored between the positivist and interpretivist paradigms – in other
words, it adopts a post-positivist approach because such a philosophy assumes that
reality is multiple, subjective, and mentally constructed by individual human beings.
“The use of flexible and multiple methods are desirable as a way of studying a small sample in-depth over time that can establish warranted assertibility as opposed to absolute truth. The researcher interacts with those being researched, and findings are the outcome of this interactive process with a focus on meaning and understanding the situation or phenomenon under examination” (Crossan 2002:48).
Furthermore, this thesis also aims to triangulate the development of
hypotheses with a deeper understanding of how customers perceive trustworthiness
and its antecedents and corollaries. In particular, in the measures used to test the
conceptual model, the research wants to incorporate the experiences of customers and
perspectives on how they rank the factors contributing to trustworthiness.
For the purpose of this research post-positivism appears to be more applicable
because it emphasises on the importance of multiple measures, each of which may
possess different types of error(s) (Grix 2004). Furthermore, this approach does admit
that the researcher cannot be entirely objective, and allows for the expression of the
researcher’s personal perspective in the research and facilitates the use of any
available instruments to achieve the research objective.
Chapter Five – Research Methodology
-108-
5.3 Research Logic
There are three logical frameworks that can be applied by the researcher; these
are inductive, deductive and abductive logic (a combination of the inductive and
deductive). In inductive logic, the researcher gathers empirical data and attempts to
form a theory based on the results of the collected data. There are no theories upon
which the researcher relies; instead, the researcher attempts to generate a theory to
explain the observed facts. In contrast, the object of deductive logic is to examine an
existing theory and find data that accommodates this theory (see Figure 5.4).
(Source: Adapted from Grix 2004)
Based on the aims and objectives of this research (see Chapter One section
1.3), the logical arguments in this thesis are deductive because they build on previous
theories surrounding trustworthiness and introduce a new model to be examined in a
specific situation. In addition it is aligned with the adopted research philosophy of the
thesis in being post-positivism, see Chapter Five section 5.2.1.2
5.3.1 Qualitative vs. Quantitative Methods
The terms quantitative and qualitative are often used to represent different
positions in research, and both methodologies are related to several different
dimensions2
2 The terms dimensions and constructs are used interchangeably in this research.
that control the process of any research (Cavana et al. 2001; Miles and
Huberman 1984). Furthermore, both methodologies can be implemented in
conjunction with other research paradigms, depending on formulation of the research
(see Table 5.2).
Inductive logic
Deductive logic
Theory
Phenomenon
Figure 5.4: Deductive and Inductive Logics
Chapter Five – Research Methodology
-109-
Table 5.2: Assumptions of the Quantitative and Qualitative Research Paradigms
(Source: Creswell 1994:5)
5.3.1.1 Quantitative Research
Quantitative research attempts to measure a phenomenon by converting it into
figures and numbers that can then be analysed using a range of statistical methods
(Collis and Hussey 2003). Furthermore, quantitative research “is often privileged as
“hard” science, a quantitative researcher relies on numbers, rates, and percentages
typically presented in a table, grid, or chart in order to communicate meaning”
(Hesse-Biber and Leavy 2004:1). Huysamen (1997) suggests that quantitative
research distinguishes between a cycle of successive stages comprising hypothesis
formulation, data collection and analysis.
Employing deductive logic, quantitative research seeks to establish facts,
make predictions, and test hypotheses; a large part of quantitative data analysis is
Chapter Five – Research Methodology
-110-
statistical (Huysamen 1997). The research is driven by the researcher’s notions about
which dimensions are of interest, and the data is thought to exist independently of the
researcher (Miles and Huberman 1984; Sarantakos 1998).
Easterby-Smith et al. (1997) suggest that in the case of the quantitative
methodology and positivist philosophy, the main advantages consist of providing a
broad coverage of the range of situations. Such an approach to research can be fast
and economical, particularly when statistics are collected from a large sample; also,
quantitative methods allow researchers to measure and control their research
variables, with the consequence that their results are statistically reliable and can be
projected from the sample to the population (Miles and Huberman 1984). There are
several types of quantitative methods; examples are experiments, quasi-experiments
and surveys.
Criticism of the quantitative methodology consist of the fact that quantitative
research is neither suitable nor cost-effective for learning why people act or think as
they do (McCullough 1995), i.e. such research is unsuitable for investigating human
actions, behaviours and attitudes. Theories resulting from quantitative studies often
fail to provide a complete and rounded view of individuals (Edwards 1998). Also, in
order to permit reliable statistical analysis the research questions must be clear,
simply stated and often open-ended, as well the sample being relatively large
(Edwards 1998). Finally, a serious disadvantage of quantitative research is that
concerns are only measured if they can be formulated prior to the beginning of the
study (McCullough 1995).
5.3.1.2 Qualitative Research
Qualitative research, in contrast to quantitative research, focuses on the
fundamental idea that it is possible to obtain knowledge about an individual’s world
through descriptive language (Sarantakos 1998), i.e. phenomena are observed and
examined by being converted into words (Baker 2001). Qualitative methods offer a
range of theoretical and methodological possibilities (Crossan 2002).
In short, qualitative methodology is the practice of qualitative research (see
Table 5.3). While quantitative research is positivist in its foundation, qualitative
Chapter Five – Research Methodology
-111-
research takes mainly a phenomenological perspective. Quantitative research consists
mainly in the collection of numerical data, but the qualitative researcher mainly
employs verbal (non-numerical) data, for instance, observations and interviews (Gay
and Airasian 1999). Furthermore, quantitative research attempts to find evidence
which supports or refutes an existing proposition, while qualitative research allows
propositions to emerge from previously existing situations (Huysamen 1997) (see
Table 5.3).
Table 5.3: Differences between Qualitative and Quantitative Research Strategies
(Source: Bryman and Bell 2003:25)
Despite the specific advantages and drawbacks of both the qualitative and the
quantitative methodologies, the application of a particular method may not be
appropriate for the purpose of this thesis because this thesis attempts to collect and
analyse data using different perspectives (questionnaire survey and in-depth
interviews). In addition, Nau (1995:47) states that “…qualitative and quantitative
methods used in conjunction may provide complementary data sets which together
give a more complete picture than can be obtained using either method singly.”
Consequently, this research will look to the theory of triangulation in order to choose
the appropriate methodological foundation for this research.
Chapter Five – Research Methodology
-112-
5.3.1.3 Combined Methods/Triangulation
The use of multiple research methods is called triangulation (Easterby-Smith
et al. 1997), and it is often employed because “no single method will ever meet the
requirements of interaction theory” (Denzin and Lincoln 1998:25). The term is used
in surveying and cartography, where “a minimum of three reference points are taken
to check an object’s location” (Easterby-Smith et al. 1997:133). Abrahamson (1983)
suggested that mixing methods prevents research becoming method-bound. In
addition, the strength of the research is increased by the counterbalance of qualitative
and qualitative methods; i.e., mixing methods avoids disadvantages of both the
qualitative and qualitative methods (Collis and Hussey 2003) (see Table 5.3).
There are four categories for triangulation research; these can be summarised
as:
• Theory triangulation: the adaptation of existing models from different fields,
and employing them in the current research area (Denzin and Lincoln 1998).
• Investigator triangulation: where several researchers collect data in the same
area, and then compare all their results. This triangulation grants all the
researchers a deeper insight, bringing different perspectives to the same
problem.
• Data triangulation: where data is collected over different time scales or from
different sources (Collis and Hussey 2003).
• Methodological triangulation: where both quantitative and qualitative
methods are applied during the same research. There are various methods to
achieve this, including questionnaires, interviews, telephone surveys and field
studies (Easterby-Smith et al. 1997).
Chapter Five – Research Methodology
-113-
Table 5.4: Different Research Methodologies
(Source: Creswell 1998:17)
To summarise, qualitative and quantitative research attempts to examine
phenomena by applying different research methods (see Table 5.4). These
methodologies are divided according to whether the researcher’s assumptions are
thought of as being subjective or objective (Denzin and Lincoln 1998). By mixing
methodologies the research captures both perspectives; the subjective perspective will
provide a full explanation of the researched phenomenon, while the objective
perspective will allow a further exploratory examination of the phenomena. The
results may be understood both in terms of the statistical and numerical (quantitative)
data, and in terms of the descriptive (qualitative) data.
This study adopts a multi-methodological perspective by conducting semi
structured interviews and a survey questionnaire; this is discussed in detail in section
5.4.1 and 5.4.2. According to Denzin (1970), triangulation is a methodological
combination, or it is a research design that combines dissimilar methods (both
qualitative and quantitative) to measure the same phenomenon (Denzin and Lincoln
1998). The use of triangulation research is advocated by LeBlanc (1996), who
suggests that such research provides greater reliability of results. Also, Denzin and
Lincoln (2000) argue that the use of the triangulation approach enhances the validity
and reliability over and above that for a single methodological approach for the same
research.
Chapter Five – Research Methodology
-114-
5.4 Data Collection
The data collection for any research passes through several steps in order to
build a valid and reliable method (see Figure 5.5). Furthermore, data collection
consists mainly of two types, resulting in secondary and primary data. Secondary data
refers to information that already has been documented about a certain phenomenon,
but which has not been gathered primarily for the researcher’s specific study
(Creswell 1994). Primary data, on the other hand, is information that has been
collected from the original source for the researcher’s particular study (Creswell
1998).
The secondary data in this research consists of materials related to scientific
research publications from established journals and books. Sarantakos (1998) stated
that it is essential to have a critical attitude towards this type of data, since the
researcher has not collected it and the precise collection process is, therefore, not
exactly replicable (Aaker et al. 2001).
(Source: Collis and Hussey 2003:152)
Figure 5.5: Data Collection Process
Chapter Five – Research Methodology
-115-
The process of collecting primary data for any research depends on the nature
of both the researcher and the phenomena under investigation. For instance, if the
researcher intends to collect in-depth information from a small sample of people,
especially concerning their emotions, experiences and sensations, conducting
interviews are a good option, but if the researcher seeks to generalise the findings,
questionnaires are more suitable (Collis and Hussey 2003).
For the purpose of this research it is necessary to gain a reflective insight
regarding relationship marketing and its association with trustworthiness from the
customers’ perspective, as well as to test the proposed model. Therefore, this research
uses semi structured interviews to refine the generated pool of items (see chapter Six
section 6.5), followed by a survey questionnaire to allow the collection of a large
sample in order to facilitate the generalisation of the proposed model as stated in the
first aim of the thesis (see chapter One section 1.3). Grix (2004) contends that
surveys are the main form of primary research undertaken, and suggests that the
technique’s popularity is due to the following factors:
1. The objectives of most research require factual, attitudinal and/or behavioural
data. Survey research provides the researcher with the means of gathering
qualitative and quantitative data, both of which are required to meet such
objectives.
2. One of the greatest advantages of survey research is its scope: a great deal of
information can be economically collected from a large population.
3. Questionnaire survey confirms to the specifications of scientific research: it is
logical, deterministic, general, parsimonious and specific.
5.4.1 Interviews
Easterby-Smith et al. (1991) argued that interviews are often seen as the best
method for gathering information, although they are complex and this complexity is
occasionally underestimated by researchers. They are also time-consuming to conduct
and analyse, and they have sometimes been used where research should have been
conducted in other ways (Robson 2002). For example, if the researcher wants to
obtain answers to a very simple question, a questionnaire might be more appropriate.
Chapter Five – Research Methodology
-116-
In a very basic sense, an interview is a series of questions that a researcher
asks respondents in person (Macionis and Plummer 1998). However, there are several
types of interviews; an interview can be structured (the researcher asks clearly defined
and precise questions) or semi-structured (the researcher can alter the question type or
number depending on the interviewee). Floyd and Fowler (1993) also argued that
interviews can be unstructured, mainly by allowing some of the questioning to be
assumed by the responses of the interviewee. Comparatively, structured interviews are
prepared and can be shaped by closed questions, whereas unstructured interviews are
relatively unsystematic and can be shaped by open questions.
The semi-structured interviews for this thesis are designed to qualitatively
analyse respondents’ comments on the main subject without constraints (Robson
2002). It will allow exploring “… the actor’s definition and how people act which
gives meaning to their own lives” (Eyles 1989:380). It also have been used to
validate, improve, and path the way to the data collection for the survey stage.
5.4.2 Survey Questionnaires
Questionnaires are central to surveys (Baker 2002:185) and they are the main
used method of data collection in field research (Stone 1978). The key aim of a
questionnaire is to collect a large sample of people’s views (Stroh 2000); therefore,
they can be used to achieve a broad image of the factors influencing customers’
decisions regarding a certain incident (House 1985). From a statistical perspective, the
large sample size needed for questionnaire studies is designed to produce unbiased
statistical results, which can then be implied for the whole population (Easterby-Smith
et al. 1991; Robson 2002). Additionally, other advantages of questionnaires are: cost
efficiency when working with large sample sizes (Robson 2002); simplicity of
analysis; and effectiveness as a way of decreasing researcher bias. Data collection and
analysis can even be completed using computer software packages (Easterby-Smith et
al. 1991; House 1985).
Survey questionnaires often examine the preferences, attitudes, practices,
concerns, or interests of a particular group of people (Gay and Airasian 1999).
Surveys can also include cross-sectional and longitudinal studies using questionnaires
or interviews for data collection, with the aim of assessing the features of a large
Chapter Five – Research Methodology
-117-
sample of interest based on a smaller sample from that population (Easterby-Smith, et
al. 1997).
In most cases, survey research allows the researcher to obtain data about
practices, situations or views at a specific point in time through conducting
questionnaires or interviews (Robson 2002). Creswell (1994) argued that the
advantages of surveys include: economy of design; the rapid turnaround in data
collection; and the ability to identify attributes of a population based on examination
of a smaller group of individuals.
Alreck et al. (1985) argues that the most important advantages of surveys are
that they are comprehensive, customised, versatile, flexible and efficient. Interpretive
surveys are employed in specific circumstances where the respondent is asked to
describe what they do, how, when, where and so on (Robson 2002). Interpretive
surveys are often described as the first step in primary data collection when the
researcher is attempting to gain an insight about the topic under investigation (Robson
2002). In-depth interviewing and focused group interviews are widely used, often to
define the questions which will later be included in a formal questionnaire for use in a
factual or opinion survey (Denzin and Lincoln 1998).
The reason behind adopting a survey questionnaire in this thesis is because a
survey provides one way of obtaining and validating knowledge. Moreover, surveys
are a means of going from observation to theory validation, a process that has three
purposes: description, explanation and exploration. This research is aiming to cover a
large population in the empirical stage of the data collection; therefore, it would
achieve significant benefit from the use of survey questionnaire.
Chapter Five – Research Methodology
-118-
5.5 Sampling
A sample is expressed as a small cluster of units selected from a larger cluster,
known as the population (Robson 2002). A sample is defined as: “…the entire group
under study as specified by the objective of the research” (Pedhazur and Schmelkin
1991:63). A population refers to “a body of people or to any other collection of items
under consideration for research purposes” (Collis and Hussey 2003:155). By
investigating a selected sample, it is expected that the researcher will be able to
present valid results and conclusions that will allow either a general view of the entire
population (if the researcher is following a quantitative method), or an in-depth view
about the selected sample (if the researcher is following a qualitative method). There
are many different sampling methods, but a good sample should always be randomly
selected, unbiased and large enough to satisfy the needs of the investigation being
undertaken (Collis and Hussey 2003).
Prior to collecting the sample, it is important that the researcher defines the
population, including a description of the members to be included (Czaja and Blair
1996). Nevertheless, there are several methods of sampling and ways to determine the
appropriate sampling size for any research. The most common methods of sample
selection are random and non-random sampling.
Non-random sampling is widely employed as a method for case selection in
qualitative research, and also in quantitative studies of a preliminary and exploratory
nature where random sampling is impossible or too expensive. However, random
sampling techniques are always strongly preferred, because they allow statistical
conclusions to be drawn (Robson 2002). In particular, there is no technique to
measure the validity of the results from non-random samples.
Random sampling relates to data collection where each member in the
population has a pre-determined possibility of being selected for inclusion in the
sample. In general this is an equal chance of being selected. There are various types of
random sampling, such as: simple random sampling, equal probability systematic
sampling, stratified simple random sampling, and cluster sampling.
Chapter Five – Research Methodology
-119-
Simple random sampling was chosen as the most appropriate method for this
research, since customers and hotels were approached randomly without any prior
knowledge or expectations about their behaviours or attitudes. The population in this
study was defined as all hotel customers. All customers from four- and five-star hotels
in Jordan were selected as a sampling frame.
The selected hotels3
In summary, a total of 556 responses were collected for the main study
were targeted after discussing the nature of the research
with the Jordanian Hotel Association; they facilitated contact with the hotels. In order
to obtain management approval for the research, initial contact was through the
Human Resources Departments in the respective hotels. Consent was given for the
researcher to conduct the research and contact customers directly in order to distribute
the questionnaire in the hotel’s lobby. Also, three assistant researchers were recruited,
whose role was to facilitate questionnaire completion and ensure the usability of the
questionnaires. The assistants were paid a daily fee of £50, and the research covered a
period of three days, from 8am to 6pm.
5.5.1 Sample Size
Hair et al. (2006) recommended obtaining a sample of three to five hundred in
order to be sufficient for data analysis, particularly when applying multivariate
analysis techniques. Therefore, respondents were approached until the sample reached
the target of five hundred completed questionnaires (n=556 to be exact). Obtaining
the recommended number of questionnaires ensured a ratio of fifteen responses to
each item in the questionnaire; the questionnaire contains thirty-six items (Tabachnick
and Fidell 2006). Moreover, other studies with similar levels of complexity used
comparable sample sizes (see for example, Anderson and Weitz 1989; Moorman et al.
1992; Morgan and Hunt 1994; Mayer et al. 1995; Sirdeshmukh et al. 2002; Caldwell
and Clapham 2003).
4
3 The participated hotels did not want to be identified. 4 The pilot study had its own sample and will be discussed in Chapter Six.
. The
sample was used to validate the measurement model and test the research hypotheses.
A detailed analysis of the questionnaires and how the data was coded and then
analysed is contained in the analysis chapters (chapters six, seven and eight).
Chapter Five – Research Methodology
-120-
5.6 Measurement Development
This section outlines the strategy used to obtain robust and reliable measures
for the research instrument. The American Psychological Association (1995) stated
that research measures should provide strong evidence of content validity, criterion-
related validity, construct validity, and internal consistency (as will be discussed later
on in this chapter in details). In order to achieve these criteria, several steps were
followed to build a robust scale and a strong measurement instrument to test the
proposed model, these are summarised in Table 5.5.
Chapter Five – Research Methodology
-121-
Table 5.5: Breakdown of the Research Approach
Step Analytical technique(s) used Purpose Specify domain of construct
Literature search Conceptualisation based on extensive exploratory study
To clearly delineate the area of the construct, so as to determine what ‘belongs’ to the construct and what does not.
Generate sample of items
Review of existing scales Review of exploratory study transcripts Item editing
To generate a large pool of items for each construct (90 items)
Expert judging (3 judges)
Sorting task (judges asked to allocate items to the construct they believed it tapped)
To select items with the highest level of agreement among experts (Rossiter 2002)
Questionnaire pilot (60 respondents)
Inter-item correlations Exploratory Factor Analysis
To take out items which displayed poor item-total correlations or loaded poorly on the expected factor
Collect data (n=556)
Measure purification – internal consistency
Exploratory Factor Analysis Item-total correlations
EFA, load significantly on the same factor
Assess normality Study of kurtosis and skewness To ensure data suitable for Structural Equation Modelling
Verify unidimensionality
Confirmatory Factor Analysis (examination of residuals)
To weed out any items displaying high residuals with other items on their factor or with items on other factors
Assess reliability
Confirmatory Factor Analysis
Having weeded out items with poor reliability, to ensure that the Cronbach alpha reliability test of each factor is acceptable >.7 (Hair et al. 2006)
Assess face validity
Relate remaining items back to the construct definition
Assess convergent validity
CFA – to ensure that all factor regression coefficients are statistically significant and substantial, in a well-fitting mode
Chapter Five – Research Methodology
-122-
5.7 The C-OAR-SE Procedure for Scale Development
There are several scale development procedures in the field of marketing.
These scales have been developed to measure various attitudes, perceptions, or
opinions of customers/consumers in various organisational settings in order to
examine a set of hypothesised relationships with other constructs or behaviours
(Netemeyer et al. 2003). Churchill’s (1979) pioneering data driven model was the
most common approach for over two decades. However, this model has been
criticised by several authors; for instance, Flynn and Pearcy (2001) noted three main
problems with the model:
1. The sample size used in scale development;
2. The fact that Churchill’s (1979) procedure contains insufficient replications;
and,
3. Issues surrounding the problem of when exploratory factor analysis is
appropriate (regarding theoretical vs. applied scales) (Flynn and Pearcy 2001).
Rossiter (2002) proposed, what he claims, a paradigm shift in the field of
marketing with the development of a new measurement scale, the C-OAR-SE
procedure for scale development. This new scale is an attempt to overcome the
limitations inherent with Churchill’s (1979) procedure.
The C-OAR-SE procedure consists of a six-fold model. The aspects of the
model are:
(C) Construct definition.
(O) Object classification.
(A) Attribute classification.
(R) Rater5
(S) Scale formation.
identification.
5 Rater(s) is a term Rossiter (2002) uses to describe respondents; this originated from the idea that respondents ‘rate’ the scale items in the questionnaire.
Chapter Five – Research Methodology
-123-
(E) Enumeration and reporting.
This model consisted of a scale development procedure and mainly based on
one type of validity (content validity). Rossiter’s (2002) approach criticised much of
the existing scale development literature in marketing, by arguing against the logic
behind adopting and following the classic procedure for developing scales (for
instance, Churchill 1979).
This thesis mainly adopts the C-OAR-SE procedure in order to develop a
robust scale to test the proposed model; this procedure will allow increasing the
content validity of the proposed model over relying entirely on data analysis for
model validation. However, it is also necessary to take on board the criticism of
Rossiter’s (2002) work particularly in relation to items selection. Hence, all the
traditional scale development procedure steps (for example, exploratory factor
analysis tests) were also considered, in order to add further validity to the process. It
is also worth noting that there is a great emphasis in this thesis on content validity.
After an examination of Rossiter’s (2002) rationale of why the C-OAR-SE procedure
is effective and valid, the research in this thesis was designed to follow the new
process for reasons that will be outlined in section 5.7.7.
5.7.1 The Construct Definition
The construct definition is defined according to Edwards and Bagozzi
(2000:156) as: “...a conceptual term used to describe a phenomenon of theoretical
interest.” However, the constructs should be defined and illustrated in terms of: (1)
the object, including its components; (2) the attribute, including its components; and
(3) the rater entity (Rossiter 2002:308). Failure to conceptualise these components
will result in insufficiency concerning how the construct should be operationally
measured, and will therefore lead to misleading definitions (Rossiter 2002).
5.7.2 Object Classification
The object classification in the C-OAR-SE procedure can have three forms: a
singular form, (which will require just a single-item part); a collective of constituents
(multiple-item parts identifying the main constituents); or multiple components (an
Chapter Five – Research Methodology
-124-
abstract form). Rossiter (2002) illustrates the object classification differently; these
classifications are:
a) Concrete singular objects: contrasts with the assumed understanding in
traditional scale development procedures, where assumption is that almost all
raters know what the object is, and, hence, there is only one object. When
applying C-OAR-SE, a group of expert judges is used to approve the
classification of the object (Rossiter 2002).
b) Abstract collective object: objects that are mixed from the raters’ perspectives;
they are seen as separate ingredients, but form a set at a higher categorical
level in the eyes of the researcher (Rossiter 2002).
c) Abstract formed object: this type of object occurs in general when people’s
explanations of the object are in opposition, or where people perceive the
object as containing different components. An abstract formed object is not
simply a collection of concrete objects, and, therefore, it is not an abstract
collective object (Rossiter 2002).
In the current research, measures were developed for each of the seven
constructs in the proposed model6
The third component in C-OAR-SE is the classification of the attribute in each
construct, i.e. the dimension on which the construct is to be judged (Rossiter 2002).
Attribute classification is generally seen as the most complicated step in the C-OAR-
. These include: consistency, competence, integrity,
benevolence, value alignment, communications (the independent variables) and
trustworthiness (the dependent variable). Items from published scales were
augmented with items developed from the pilot study, and the measures were
developed, purified and validated following a rigorous process (graphically
represented in Chapter Six). Definitions for the nine constructs were presented in
Chapter Four Table 4.3 and section 4.6.1. Each construct was clearly defined, and
compared and contrasted to similar constructs in the literature. The definitions also
prompted the development of further items; consequently theoretical discriminate
validity was prevented.
5.7.3 Attribute Classification
6 See Chapter Four for more details regarding the proposed model section 4.3
Chapter Five – Research Methodology
-125-
SE scale development procedure because of the difficulty of classifying some
attributes. The three types of attribute are concrete (singular), (abstract) formed, and
(abstract) eliciting:
a) Concrete attribute: according to Rossiter (2002), several of the attributes that
are measured in marketing are concrete. As with concrete singular objects, a
concrete attribute has almost complete agreement from raters, who clearly
understand that there is only one feature being referred to when the attribute is
created and applied to the object being rated in a questionnaire or interview.
b) Formed attribute: this type of attribute consists of those that are abstract, in
which raters’ answers vary if they are asked what the characteristic is, and
formed attributes that are composite and multi-componential (Rossiter 2002).
c) Eliciting attribute: here the attribute is abstract (raters’ answers would differ
moderately if asked what the characteristic is) but in this case the attribute is
an ‘internal’ characteristic (Rossiter 2002).
5.7.4 Rater Identification
The last part of the construct definition is the rater identification, referring to
the rater by entity; however, it should be noted that constructs vary according to
whose perception they represent. Rossiter (2002) postulated three types of rater
entity, namely, individuals, experts and groups:
a) Individual raters: consist mainly of one type of rater entity, seen as an
individual. However, “the rater entity is the INDIVIDUAL for all individual
difference constructs that involve self reports” (Rossiter 2002:318). Experts
and peers may rate individual differences; however, individual self reports are
the most common type of rater entity for constructs in the marketing field
(Rossiter 2002).
b) Expert raters: these are required in some situations when the constructs
require experts to conduct the ratings. Fundamentally, the expert raters are
conducting a content analysis, and, therefore, the reliability of the ratings
depends on achieving high inter-judge agreement (Rossiter 2002:319).
c) Group raters: the third classification for the rater entity of a construct is the
group. According to Rossiter (2002), the group, in marketing, is often seen as
Chapter Five – Research Methodology
-126-
a sample of consumers or industrial buyers, but in certain situations it can be a
sample of managers, or general employees. However, the reliability of scores
depends mainly on the group sample size (Rossiter 2002).
Rossiter (2002) argued that during the scale development procedure the
researcher has to consult with expert judges; this enables the research to benefit from
deeper insight into the research items, which allows for rigid item purification. The
exact role of the judges is explained in detailed in Chapter Six - section 6.4.
5.7.5 Scale Formation and Instrument Description
The scale formation in the C-OAR-SE procedure consists of four levels:
a) Object item parts and attribute item parts: scale formation in C-OAR-SE is a
matter of putting together object item parts with their corresponding attribute
item parts to form scale items, since the response alternatives have to be
added for each item (Rossiter 2002). The number of items required to outline
the scale remains the same regardless of whether the rater entity is an
individual, a group of experts, or a large sample. The content (wording) of the
scale items is not independent of the rater entity; thus, the items must be
easily understood by the target raters, and a pre-testing phase is required to
ensure this (Rossiter 2002:319).
b) Pre-testing scale items: as recommended by Rossiter (2002), the most
effective method for pre-testing scale items is cognitive interviewing, which
includes extensive searching and rating questions, to make certain they are
understood as planned (Schwarz 1999). However, the central step of pre-
testing items for meaning is conducted during the development of marketing
scales in academic research.
c) Response formats: Rossiter (2002) suggests that Likert response systems
should not be implemented because they cannot provide definite, precise item
scores (Rossiter and Percy 1987:547). In typical items for which Likert
responses are used, the strength of response is built into the item stem. This
thesis followed an alternative method in which it built intensity of response
into the leaves of the item (the response alternatives). For numerical scales,
which can be used for probability, frequency, or degree, five to seven
Chapter Five – Research Methodology
-127-
categories seems to best fit the number of psychological discriminations that
most consumers can make with regard to an attribute.
d) Randomised order: Rossiter (2002) postulated that with multiple-item scales,
the order of the items should be randomised to minimise response-set artefacts
in the obtained scores. This means a randomised presentation across multiple
items for objects (constituents or components) as well as for attributes (items
within components should be separated).
The main study questionnaire consisted of a three-page paper document. A
copy of the full questionnaire is shown in Appendix 1. There were five main sections
in the questionnaire, as described below:
• Section one: contained instructions on how to complete the questionnaire, with
a statement explaining the aims and objectives of the research and reassuring
respondents that their answers were anonymous and confidential.
• Section two: contained four questions to ascertain whether respondents had
visited the hotel before, what type of service(s) they had used and how
frequently they used the hotel and the hotel’s facilities.
• Section three: the main part of the questionnaire, containing the scales. This
section contained a series of statements with 5-point Likert scales to indicate
levels of agreement with the statements. The items were presented in a random
order, to avoid response bias and the false inflation of reliability, i.e.
questionnaire fatigue.
• Section four: contained demographical information about the respondents,
specifically, age, gender and nationality.
A full description of the questionnaire items is included in the discussion of
the scale development (in Chapter Six).
5.7.5.1 The Generation of the Initial Research Items
Once each of the nine constructs had been clearly defined, a sample of items
related to the constructs was developed. The aim of this stage was to capture all
aspects of a construct, while building some redundancy into the instrument (DeVellis
1991). The items were generated in four different ways:
Chapter Five – Research Methodology
-128-
• From the existing literature: items were developed on the basis of various
discussions in the literature.
• From published scales: these were gathered and individual items were
considered for adoption or adaptation. Only often-cited scales were
considered; journal ranking and frequency of citation was taken into
consideration.
• From in-depth interviews: these consisted of card-sort exercises and open
discussion with nine interviewees. This process enabled the researcher to
access the terminology favoured by actual consumers, and to generate extra
items based on their personal experience with the service provider.
• From the researcher’s personal assumptions: items were created based on the
definitions of each construct, to ensure that all aspects had been considered.
Once the pool of items had been developed, they were edited. DeVellis’
(1991) recommendations were followed concerning item length, reading difficulty
and double-barrelled items (which express more than one idea).
5.7.6 Enumeration
Rossiter (2002) argued that there are two factors to consider within
enumeration these are:
a) Indexes, averages, and single-item scores: because of the different object and
attribute types that can be combined in a construct, enumeration rules will
vary between constructs (Rossiter 2002).
b) Reporting scale scores: the enumeration rules imply that indexes will receive
absolute total scores and items for eliciting attributes will receive averaged
scores (Rossiter 2002).
5.7.7 Criticism of C-OAR-SE
The C-OAR-SE procedure is an innovative scale development procedure in
the field of marketing. In fact, at the time of writing the scale has not yet been tested
empirically or even reviewed by many scholars and academics. There do exist one or
two examples in the literature; see for instance, Diamantopoulos (2005) and Finn and
Kayande (2005). Diamantopoulos (2005) identified several problems with Rossiter’s
Chapter Five – Research Methodology
-129-
(2002) model; one of these arguments was that the procedure goes against the fact that
constructs are abstract entities by their nature (Diamantopoulos 2005:2). Bearing the
above in mind, the notion of construct definition as used by C-OAR-SE would benefit
from some clarification. Such clarification is important for two reasons:
• The degree of acceptable aggregation during construct definition according to
C-OAR-SE is not entirely clear.
• The foundation of the object classification under C-OAR-SE is not clear; in
that denotative and connotative meaning appear to be confused.
Furthermore, Diamantopoulos (2005) argues that if the denotation of an object
is open to several interpretations, then its conceptual clarity will be negatively
affected. However, in the case of concrete singular and abstract collective objects, the
classification is less clear (Diamantopoulos 2005). Finally, the distinction between
abstract collective and abstract formed objects is hazy in that the ‘constituents’
comprising the former and the ‘components’ comprising the latter are both deemed to
be concrete singular objects (Diamantopoulos 2005).
Diamantopoulos (2005) suggests that a way to solve these problems would be
to classify objects into concrete singular and abstract collective categories, but at the
same time allow that the constituents of abstract collective objects may themselves be
multi-componential (Diamantopoulos 2005).
In addition, further criticism of the C-OAR-SE procedure was provided by
Diamantopoulos (2005) in terms of attribute classification, concrete attributes, and the
fact that the fundamental relation between attribute and measure is not considered
when discussing formed attributes. Still, the main point to be noted here is the
problem of how to select a single ‘good’ item. Diamantopoulos (2005) suggested
applying the index construction procedure developed by Diamantopoulos and
Winklhofer (2001), where a ‘good’ item is one that (a) captures a particular facet of
the construct’s domain of content, (b) is not collinear with other items, and (c) its link
with the latent variable has a non-zero coefficient. This research took on board all the
criticism to the C-OAR-SE in particularly selecting the research items as it has been
discussed in Chapter Six.
Chapter Five – Research Methodology
-130-
5.8 The Structure of the Data Collection
This section covers the parameters used to ensure the robustness of the study
design, it examines the internal consistency, uni-dimensional, measurement
validation, and normality evaluation of the research items. However, the data
collection process is reported in detail in Chapter Seven. 529 usable responses were
obtained. The main study served to purify and validate all measures, as well as test the
hypothetical framework.
For the first objective, the sample of 529 responses was split into two half
samples, using the random function in SPSS. The first half sample (n=265) was used
to purify the measures and explore uni-dimensionality. The second half sample
(n=264) was used to validate the measures, verifying uni-dimensionality and
assessing reliability and validity. The analysis of the remaining two objectives was
conducted on the full sample of 529 cases.
5.8.1 Internal Consistency
Inter-item correlation tables were inspected to identify and delete items with
poor inter-item correlations. The domain sampling model rests on the assumption that
items which belong to the same domain share the same amount of “common core”
(Churchill 1979:68); consequently, items which correlate poorly with other items
cannot share the same amount of common core and cannot be related to the same
construct. Likewise, items with a low item-total correlation were candidates for
deletion. It was important to use both methods since, for example, the item-total
correlation method does not account for external consistency (Gerbing and Anderson
1988). Also, since several constructs were expected to correlate highly, items
belonging to several factors may still have displayed high item-total correlations.
Chapter Five – Research Methodology
-131-
5.8.2 Unidimensionality Exploration
Unidimensionality is a cornerstone of measurement theory and the
homogeneity of items (Netemeyer et al. 2003). As Hattie (1985: 49) points out, “that
a set of items forming an instrument all measure just one thing in common is a most
critical and basic assumption of measurement theory.” Netemeyer et al. (2003:9)
confirmed Hattie’s (1985) statement when they stated: “when the measure is
multidimensional, items tap more than one factor ... Hence, the scale used to
operationalise the construct should reflect the hypothesised dimensionality.” While
Exploratory Factor Analysis (EFA) cannot verify unidimensionality in the strict sense
of the term (Gerbing and Anderson 1988), it can be used as a first-cut test, to identify
any items tapping a separate construct.
The items which seemed to reflect the same construct were factor analysed,
factor by factor, to ensure they all loaded on the expected factor. When a second
factor was extracted, the items loading on the second factor were studied to decide
whether the factor was conceptually meaningful and was to be retained or added to
the conceptualisation. When this was not the case, then items loading the highest on
the second, unwanted factor(s) were removed and the analysis was re-run. This
process repeated until it returned a single factor.
Once this procedure had been applied to all dimensions expected to load on
the same construct (Sense-making potential and exploratory potential), the retained
items of all factors for each construct were submitted to another EFA, to ensure that
all items loaded on the expected dimension. Items which loaded on several factors, or
which did not load on any factor, were removed, since these two situations were
indications of items which did not relate to one and only one factor, thereby
preventing the measure from being both unidimensional and internally consistent.
Items which loaded on a dimension other than which was expected were reviewed
alongside the definition of the dimension they loaded on. They were either retained in
the ‘new’ dimension or removed, based on an assessment of their content validity.
Chapter Five – Research Methodology
-132-
5.8.3 Measure Validation
The validation of the measures was carried out on the half of the main study’s
sample that was not used during the item purification stage. By using this method, the
chance of basing decisions purely on a sample’s idiosyncrasies was reduced. The size
of the validation sample was 264.
5.8.4 Normality Assessment
It is crucial to ensure that there are no fundamental departures from normality.
For this purpose, the kurtosis and skewness of each item were analysed and
confirmed, showing that all measures had skewness and kurtosis statistics within the -
1.96 to +1.96 range. The boundaries of this range are the points for rejecting the
freedom from skewness or kurtosis assumptions at the .05 error level (Hair et al.
2006). Furthermore, the chosen estimation method, Maximum Likelihood, remains
robust in situations where the assumption of multivariate normality is moderately
violated, as long as the sample size exceeds 100 (Gerbing and Anderson 1985).
Chapter Five – Research Methodology
-133-
5.9 Quality of Research
Any discussion of research methods often leads to two essential topics, namely
the reliability and validity of the research. The research reliability refers to issues of
consistency and dependability, while the validity of the research consists of the
existence of an attribute (such as the characteristics of a determinant of the model)
that influences the outcome of the measurement procedure (Borsboom et al. 2004).
5.9.1 Research Validity
The validity of the research is examined by assessing the process of the
research and making sure there are no rational errors in drawing conclusions from the
obtained data (Easterby-Smith et al. 1991). However, Pedhazur and Schmelkin (1991)
distinguished between validity of the research and validation, where validity is about
ontology and validation is about epistemology. With this distinction in mind, it is
apparent that most of the validity literature has not dealt with the problem of validity
but with the problem of validation. Borsboom et al. (2004) continued this discussion
and argued that there is nothing wrong with describing, classifying, and evaluating
validation strategies. However, if one concentrates on the epistemological problems
long enough, one will move away from the validity concept rather than toward it.
Nonetheless, Messick (1989:13) defines validity as follows:
“Validity is an integrated evaluative judgment of the degree to which empirical evidence and theoretical rationales support the adequacy and appropriateness of inferences and actions based on test scores or other modes of assessment.”
There are two ways of examining the validity of the research; empirical and
theoretical validation. Empirical validity assesses the validity of the measures checked
against empirical evidence, while theoretical validation examines the theoretical
concepts that are the basis for the research (Sarantakos 1998). However, both types of
validity have to take account of threats and biases that could weaken the aim of the
research (Easterby-Smith et al. 1991).
In this thesis several types of validity are considered, namely: translation
validity, convergent validity and discriminant validity (Netemeyer et al. 2003). The
Chapter Five – Research Methodology
-134-
following sections will address each of these types of validity and how they were
addressed in this research.
5.9.1.1 Translation validity
Translation validity contains two main dimensions: content validity and face
validity. Content validity is achieved when “a measure’s items are a proper sample of
the theoretical domain of the construct” (Netemeyer et al. 2003:12). Content validity
is also defined as: “a measure is supposed to have content validity if it covers all
possible aspects of the research topic” (Sarantakos 1998:79). In content validity, the
researcher is also anxious to make sure the items measure the full implicit area
(Sheppard 1993).
Creating content validity is mainly a subjective operation that relies on the
judgment of experts. This process varies in the levels of requirement, but in general a
researcher will want a panel of experts to appraise the provisional research variables
and indicators to determine their relevance and representativeness to the research
(Sheppard 1993). The term ‘expert’, in this context, refers to someone who has
experience, knowledge of the construct, and familiarity with the context in the
research discipline and the specific field in which the research is situated (Rossiter
2002). This was achieved by having a panel of three judges examining the research
constructs and items as it’s reported in Chapter Six.
In contrast, face validity centres on the need to ensure the research instrument
is valid from the perspective of the respondent. Face validity assesses the accuracy
and thoroughness with which measures represent a phenomenon. One way of
addressing the danger of poor content validity is to access the way customers perceive
the constructs in their own ‘language’.
5.9.1.2 Construct Validity Assessment
Construct validity (of which convergent validity is a component) refers to the
closeness of the resemblance between the constructs of the theory being tested, and
the measurement of the data collected from a sample: do the data and measurement
tools chosen faithfully represent the original constructs? Moreover, construct validity
resides in the logic of items that comprise measures of social concepts (Sarantakos
Chapter Five – Research Methodology
-135-
1998). Fine construct validity should have a theoretical basis that is translated through
explicit theory (Pedhazur and Schmelkin 1991). In contrast, weak construct validity
may be characterised by a lack of theoretical agreement about its content. Construct
validity is a method of defining something, and it is related to the extent that a
researcher’s construct is at odds with the existing literature on related hypothesised
relationships using other measures.
Construct validity is usually assessed by considering convergent and
discriminant validity, and criterion and nomological validity.
Criterion-related validity, according to Netemeyer et al. (2003), relates to the
extent to which processes external to the proposed research instrument are utilised. A
formal criterion validity assessment would have required the administration of too
many additional scales for comparison with the measures being developed. An
alternative was to study the correlations between factors, to verify that the highest
correlations were between the factors expected to load on the same higher-order
construct. This was achieved in Chapter Eight – Table 8.5.
Nomological validity refers to “the degree to which predictions from a formal
theoretical network containing the concept under scrutiny are confirmed” (Netemeyer
et al. 2003:13). In other words, nomological validity examines the degree to which the
constructs of the research which are theoretically linked are also empirically related.
Hence, nomological validity is primarily empirical because its validity is based on
investigations and constructs in terms of formal hypothesis derived from theory, and
require statistical techniques to assess it. Therefore, nomological validity was tested
using AMOS programs as been recommended by various authors (for example,
Anderson and Gerbing 1988; Bollen 1989; Schmucker and Lomax 2004).
Specifically, nomological validity was assessed when the full structural path model
was estimated7
Convergent validity assesses the extent to which each item ‘contributes’ to the
meaning of the construct being measured. A weak condition for convergent validity is
that the regression coefficient of each item loading on a particular latent variable (or
dimension) is statistically significant. A more powerful condition is that these
.
7 See Chapter Eight for a full discussion.
Chapter Five – Research Methodology
-136-
coefficients are substantial. Both conditions need to be met under acceptable model fit
(Steenkamp and van Trijp 1991). Convergent validity refers to the principle that the
indicators for a given construct should be at least moderately correlated among
themselves (Embretson 1983). However, Cronbach’s alpha is broadly associated with
convergent validity. Convergent validity was tested in this thesis in Chapter Seven,
Section 7.6.
5.9.1.3 Discriminant Validity Assessment
Discriminant validity concerns the principles that the indicators for different
constructs should not be so highly correlated as to lead one to conclude that they
measure the same phenomenon. Some researchers use the norm that the correlations
testing convergent validity should be higher than those testing discriminant validity,
based on the rationale that items measuring the same thing should correlate more
highly among themselves than with other things. However, exploratory analysis is
commonly used to establish discriminant validity (Borsboom et al. 2004).
Discriminant validity assesses whether the newly-developed scales measure
something different from other scales. Discriminant validity in this thesis was
approached in two ways. First, each construct was assessed against every other
dimension using a set of nested models. For each pair of dimensions, two models
were tested in succession, where the correlation between the dimensions was first left
free to vary, and then set to 1. Since the two models were nested (the second being a
special case of the first), a Chi-Square difference test could statistically assess
whether the correlation between the two constructs was different from 1, i.e. whether
there was discriminant validity between the two dimensions.
Second, a confidence interval of +/- 2 standard deviations around each
correlation was calculated. Any confidence intervals of 1 would have indicated a lack
of discriminant validity.
In summary, in this thesis all the various forms of validity were important. In
particular, content validity was considered to be especially significant during the scale
development process because the procedure followed Rossiter’s (2002) model that
focused on content validity over external validity.
Chapter Five – Research Methodology
-137-
Since the current research relies on the C-OAR-SE procedure, in line with
Rossiter (2002) the validity of the research relies mainly on its content validity.
However, other validity tests were also considered as will be reported later on.
Furthermore, the reliability of the results arising from the survey questionnaires were
tested using the Cronbach’s alpha reliability and composite reliability to measure the
internal consistency of the research items. However, the “content validity of the scale
must be convincingly established before precise scores can be taken to mean what
they are supposed to mean” (Rossiter 2002:328). Tabachinck and Fidell (1983)
suggest that data defects may lead to a mistaken analysis, and warn that examining
data for these defects is a prerequisite for mature analysis (Sarantakos, 1998).
Therefore, problems surrounding missing data, outliers, multi co-linearity and
violations of statistical assumptions were identified and corrected before applying
statistical procedures.
Additionally, the reliability of the research consisted of two broadly accepted
tests, examining temporal stability and internal consistency (tested using Cronbach
alpha test). This discussion will be the research foundation to allow us to proceed and
examine the scale development process.
5.9.2 Reliability Assessment
Reliability represents the stability and dependability of the research.
Reliability of the research is defined as: “the extent to which it is free from random
error components” (Judd et al. 1991:51). The necessity of stability does not imply that
all data must be stable in the sense that over an extended period a researcher will get
the same answers to the same questions (Easterby-Smith et al. 1991). However,
stability must be illustrated within the theoretical framework within which the study is
situated. The reliability test should permit only a small margin of variation between
sets of data (Cronbach 2004).
Consequently, the selection of tests for reliability must be made on the basis of
meeting the data demands of the assumptions. There is a relationship between the
concept of stability, the reliability test and the theoretical framework. Particular
procedures for examining reliability can be found in valuable sources such as
Cronbach (1954), Guetzkow (1950), and Cronbach (2004). However, there are four
Chapter Five – Research Methodology
-138-
main methods to test the reliability of research: the test retest method; the split-halves
method (Brennan 2001); the internal consistency method; and inter-rater reliability,
which assesses the degree to which different raters give consistent estimates of the
same phenomenon (Bryman and Bell 2003). There are various models for reliability
testing employed by most researchers; these are Cronbach’s Alpha, Split-half,
Guttman, Parallel and Strict Parallel reliability testing8
In the case of this thesis, temporal stability is concerned with the stability of a
respondent’s item responses over a period of time (Netemeyer et al 2003). In contrast,
internal consistency in this thesis is assessed using Cronbach’s widely accepted
coefficient alpha (α) as well as composite reliability, which evaluates the average
correlation between items in a test and is commonly used in marketing methodologies
(DeVellis 2003).
.
In a marketing sense, reliability is the extent to which a variable or set of
variables are consistent with what they intended to measure (Hair et al 1998:90). This
essentially means that reliability deals with the proportion of variance attributed to the
true score of the latent variable (DeVellis 2003:27). Within the psychometric
literature relevant to marketing there are two main forms of reliability that need to be
considered, namely temporal stability and internal consistency (Netemeyer et al
2003).
8 For more information about each type of reliability test, see Netemeyer et al. (2003).
Chapter Five – Research Methodology
-139-
5.10 Methods of Data Analysis
The issue of data analysis is concerned with ensuring that the appropriate
analysis is carried out. Multivariate methods have been used for the research in this
thesis.
Scientific enquiry is an iterative learning process; the object of research, in
this case the explanation of social phenomena, must be specified and then evaluated
by gathering data. Because of the fact that data collection normally includes
simultaneous measurements of numerous variables, there is a need to complete
multivariate analyses, i.e. multivariate analysis refers to any statistical method
employed to analyse a data set that contains more than one variable (Byrne 2001; Hair
et al. 2006). The main analysis method that this thesis relied upon is Structural
Equation Modelling, as it will be discussed in details in the following section.
5.10.1 Structural Equation Modelling
Structural Equation Modelling (SEM) is a prevailing multivariate analysis
technique, also called simultaneous equation models. SEM’s are divided into two
parts, i.e. a measurement model and a structural model. The measurement model deals
with the relationship between measured variables and latent variables. In contrast, the
structural model covers the relationships between latent variables only. One of the key
advantages of SEM is that latent variables are free of random error.
The significance of SEM for the current thesis emerges from the fact that SEM
is a general, primarily linear, cross-sectional statistical modelling technique (Byrne
2001). Factor analysis, path analysis and regression are all contained within SEM, and
furthermore, SEM is mainly a confirmatory, rather than an exploratory, technique.
Explicitly, a researcher is more likely to apply SEM to determine whether a certain
model is valid, rather than using SEM to discover a suitable model, although SEM
analyses do often involve an exploratory aspect.
The following sections outlines the assumptions and choices made in relation
to the SEM method employed to validate the empirical research measures and test the
propositions and hypotheses of this research.
Chapter Five – Research Methodology
-140-
5.10.1.1 SEM vs. Regression
All SEM techniques share two main traits: (1) they estimate several
interrelated dependence relationships; and (2) the relationships can include
unobserved phenomena, for which measurement error is taken into account during the
estimation (Hair et al. 2006). The latter characteristic gives SEM an advantage over
multiple regression, because regression estimates may be biased since they do not
take account of measurement error.
Furthermore, the multiple regression method allows for the estimation of
direct effects on only one dependent variable at a time, whereas with SEM the
researcher can estimate relationships between several independent and several
dependent variables simultaneously (Hoyle 1995). These two points are of particular
importance for this study, since the conceptual model consists of unobserved
constructs (including two higher-order constructs which, in essence, are two steps
removed from being observed constructs), and two of these (behavioural and
attitudinal loyalty) are dependent variables. Therefore, SEM was chosen over multiple
regression because it enables a more rigorous and stringent testing of the entire
nomological network in one simultaneous estimation.
This thesis applied SEM because other multivariate modelling techniques fall
short in terms of fully addressing and assessing the proposed model. For example,
multiple regression analysis deals only with one dependent variable at a time without
calculating latent constructs and error scores (Hair et al. 2008). MANOVA only
compares the mean scores of several variables without providing any indication of the
regression weight of these scores; while discriminant analysis does not take into
account error scores (Hair et al. 2008). Hence, SEM is found to be more relevant
because it examines all the constructs in the model simultaneously and calculates their
error scores.
5.10.1.2 SEM: Assumption Criteria
SEM grows out of and serves purposes similar to multiple regression, but in a
more powerful way which takes into account the modelling of interactions. This
includes:
Chapter Five – Research Methodology
-141-
• Nonlinearities;
• Correlated independents;
• Measurement error;
• Correlated error terms;
• Multiple latent independents (each measured by multiple indicators), and,
• One or more latent dependents (also each with multiple indicators).
SEM may be used as a more powerful alternative to multiple regression, path
analysis, factor analysis, time series analysis, and analysis of covariance. That is,
these procedures may be seen as special cases of SEM, or, to put it another way, SEM
is an extension of the general linear model (GLM) of which multiple regression is a
part.
The advantages of SEM compared to multiple regression include more flexible
assumptions (particularly allowing interpretation even in the face of
multicollinearity). This includes:
The use of confirmatory factor analysis to reduce measurement error by
having multiple indicators per latent variable.
The attraction of SEM’s graphical modelling interface, the desirability of
testing models overall rather than coefficients individually.
The ability to test models with multiple dependents, the ability to model
mediating variables.
The ability to model error terms, the ability to test coefficients across multiple
between-subjects groups.
The ability to handle difficult data (time series with auto correlated error, non-
normal data, incomplete data and so on).
As said earlier, SEM is usually viewed mainly as a confirmatory rather than an
exploratory procedure and uses one of three approaches:
1. Strictly confirmatory approach: A model is tested using SEM goodness of-fit
tests to determine if the pattern of variances and covariances in the data is
consistent with a structural (path) model specified by the researcher. However,
Chapter Five – Research Methodology
-142-
as other unexamined models may fit the data as well or better, an accepted
model is only a not disconfirmed model.
2. Alternative models approach: One may test two or more causal models to
determine which has the best fit. There are many goodness-of-fit measures,
reflecting different considerations, and usually three or four are reported by
the researcher. Although desirable in principle, this alterative models approach
run into the real-world problem that in most specific research topic areas, the
researcher did not find in the literature two sufficiently well developed
alternative models to test.
3. Model development approach: In practice, much SEM research combines
confirmatory and exploratory purposes: a model is tested using SEM
procedures is found to be deficient, and an alternative model is then tested
based on changes suggested by SEM modification indexes. This is the most
common approach found in the literature. The problem with the model
development approach is that models confirmed in this manner are post-hoc
ones that may not be stable (may not fit new data, having been created based
on the uniqueness of an initial dataset). Researchers may attempt to overcome
this problem by using a cross-validation strategy, under which the model is
developed using a calibration data sample and then confirmed using an
independent validation sample. This approach was used in this research. A full
discussion of the approach with the related results is included in Chapter
Eight.
The use of SEM relies on a number of assumptions: (1) independent
observations; (2) linearity of all relationships (Hair et al. 2006); (3) distributional
normality of the data; and (4) continuous data. These assumptions are now reviewed
in the context of the study. Firstly, the assumption of independent observations was
met since the design of the study precluded the collection of more than one
questionnaire per respondent.
Secondly, much of marketing research is based on the assumption of linear
relationships. Some relationships used in marketing clearly are not linear, such as the
relationship between arousal and hedonic value (Tabachnick and Fidell 2007).
However, in the absence of evidence to the contrary, marketing researchers
Chapter Five – Research Methodology
-143-
commonly assume that the relationships they hypothesise are linear. The same
assumption was made here.
Thirdly, normality assumptions were considered, testing in particular for the
skewness and kurtosis of the distribution of the variables in the SEM (Tabachnick and
Fidell 2007). The results of these tests (reported in Chapter Seven) show that all
indicators were within an acceptable range from the ideal 0, and, therefore, the data
approximates a normal distribution.
Fourthly, although strictly speaking Likert scales (which were the only
measures used to test the model) are ordinal and, therefore, cannot be continuous, they
are assumed to be categorised reflections of an underlying continuous variable
(Jöreskog and Sörbom 1996). Ordinal variables with at least five categories which do
not depart widely from normality have been found to be suitable for SEM (Hair et al.
2006).
5.10.1.3 Measurement Model
Specifically, there are two distinct components in SEM: 1) the measurement
model; and 2) the SEM. The measurement model is the component of the general
model in which latent constructs are prescribed.
The latent constructs are unobserved variables implied by the covariances
among two or more observed indicators (Hoyle 1995). By using confirmatory factor
analysis for the measurement model, a priori hypotheses regarding relationships
among and between observed indicators and their underlying latent constructs are
evaluated. Thus, the measurement model specifies the posited relationships between
the observed indicators and the latent constructs, while describing the freedom of
random error and uniqueness associated with their indicators.
According to Anderson and Gerbing (1988), confirmatory measurement
models should be evaluated and re-specified before measurement, and structural
equation models should be examined simultaneously to allow the assessment of the
overall model fit before making any assumptions on the proposed model. Thus, before
testing the overall measurement model, each construct in the model was analysed
separately. Furthermore, when each construct had an acceptable fit based on the fit
Chapter Five – Research Methodology
-144-
indices, a pair of constructs was evaluated in order to confirm that the theoretically
pre-specified variables or indicators do in fact measure what is believed to be their
underlying construct. The model is modified so that the final model becomes
theoretically meaningful as well as being statistically acceptable. This also ensures
that the final model represents the theoretical model of interest for the study. After
assessing the overall model, the psychometric properties of each latent construct were
evaluated separately by examining the completely standardised loading, the error
variance, the construct reliability, and the variance extracted.
5.10.1.4 Structural Model
The structural model is the hypothetical model that prescribes relationships
among latent constructs and observed variables that are not indicators of latent
constructs (Hoyle 1995). Generally, this model is known as the component of a
general model that relates the constructs to other constructs by providing path
coefficients (parameter values) for each of the research hypotheses. Specifically, each
estimated path coefficient can be tested for its respective statistical significance in
terms of the hypothesised relationships, while including standard errors and calculated
t-values (Bollen 1989; Byrne 1998; Hair et al. 2006).
In the structural model, a specific structure between latent endogenous and
exogenous constructs must be hypothesised, and the measurement model for latent
endogenous and exogenous constructs must be determined (Hair et al. 2006).
Commonly, maximum likelihood (ML) or generalised least squares (GLS) techniques
are utilised for model estimation because these methods allow for the analysis of
models involving latent constructs and non-zero error covariances across structural
equations (Kline 1998). If a relationship can be specified in terms of directions, a one-
tailed significance test can be employed. Otherwise, a two-tailed significance test
must be used in cases where there exists an unknown direction for a pre-specified
relationship. If the reported p-value is greater than a certain critical value, the null
hypothesis that the associated parameter is equal to zero can be rejected. This p-value
is determined by dividing the appropriate coefficient by its standard error. In general,
if an estimated t-value is greater than 1.96, the parameter indicates statistical
significance for a two-tailed test at the .05 level of significance (Tabachnick and
Fidell 2007). The coefficient is significant at the .01 level if the p-value exceeds .05.
Chapter Five – Research Methodology
-145-
These critical values (the beta and gamma coefficients) are utilised and evaluated for
testing relationships between the constructs.
As another evaluation of the structural model, the standardised solution, where
the estimated coefficients all have equal variances and a maximum value of 1.0, must
be examined (Bollen 1989; Hair et al. 2006). In order to measure the entire structural
equation, an overall coefficient of determination (R2) must be calculated to explain the
variance. As a result, the structural model provides a meaningful and parsimonious
explanation for observed relationships within a set of measured variables (Tabachnick
and Fidell 2007). The model also enables explanations of direct, indirect, and total
structural effects of the exogenous latent constructs on the endogenous constructs.
5.10.2 Confirmatory Factor Analysis
Confirmatory factor analysis (CFA) seeks to determine whether the number of
factors and the loadings of measured (indicator) variables on them conform to what is
expected on the basis of pre-established theory. Indicator variables are selected on the
basis of prior theory and factor analysis is used to see if they load as predicted on the
expected number of factors. The researcher’s priori assumption is that each factor is
associated with a specified subset of indicator variables. A minimum requirement of
confirmatory factor analysis is that one hypothesizes beforehand the number of
factors in the model, but usually also the researcher will posit expectations about
which variables will load to on which factors (Kim and Mueller 1978b:55). There are
two approaches to confirmatory factor analysis:
1. The Traditional Method: CFA can be accomplished through any general-
purpose statistical package that supports factor analysis. Note that for SEM
CFA one uses principle axis factoring (PAF) rather than principle components
analysis (PCA) as the type of factoring. This method allows the researcher to
examine factor loadings of indicator variables to determine if they load on
latent variables (factors) as predicted by the researcher’s model. This can
provide a more detailed insight into the measurement model than can the use
of single-coefficient goodness of fit measures used in the SEM approach. As
such, the traditional method is a useful analytic supplement to the SEM CFA
approach when the measurement model merits closer examination.
Chapter Five – Research Methodology
-146-
2. The SEM Approach: CFA can mean the analysis of alternative measurement
(factor) models using a SEM package such as AMOS. While SEM is typically
used to model causal relationships among latent variables (factors), it is
equally possible to use SEM to explore CFA measurement models.
This is done by removing from the model all straight arrows connecting latent
variables, and adding curved arrows representing covariance between every pair of
latent variables. The straight arrows from each latent variable to its indicator
variables, as well as the straight arrows from error and disturbance terms to their
respective variables, are all left in the model. Such a measurement model is run and
evaluated like any other model, using goodness of fit measures generated by the SEM
package. By using SEM in this thesis, the study explored CFA models with the
assumption of certain correlations among the error terms of the indicator variables.
Such measurement error terms represent causes of variance due to unmeasured
variables, as well as random measurement error.
For data analyses, in this research AMOS9
5.10.3 Statistical Methods for Hypotheses Testing
software package was used to
perform confirmatory factor analysis of the measurement items that were used to
capture the items of the constructs. Using AMOS for confirmatory factor analysis
provided a rigorous assessment of the fit between the collected data and the
theoretical factor structure, and satisfied the minimum requirements for assessing the
measurement properties.
The properties of the nine research constructs and the nine hypotheses in the
proposed structural model were tested using AMOS 8.0, and the maximum likelihood
(ML) method as an estimation technique for model evaluation and procedures (for
details, see Anderson and Gerbing 1988; Bentler 1983; Byrne 1998). The one stage
testing processes was also utilised.
SEM is designed to evaluate how well a proposed conceptual model
containing observed indicators and hypothetical constructs explains or fits the
collected data (Bollen 1989; Hoyle 1995). It also provides the ability to measure or 9 The reason of using AMOS software is because it’s the standard package provided by Coventry University.
Chapter Five – Research Methodology
-147-
specify the structural relationships among sets of unobserved (latent) variables, while
describing the amount of unexplained variance (Byrne 1998; Hoyle 1995).
Clearly, the hypothetical model in this study was designed to measure
structural relationships among the unobserved constructs that are set up on the basis
of relevant theories and prior empirical research and results. Therefore, the SEM
procedure is an appropriate method for testing the proposed structural model and
hypotheses for this study.
According to Byrne (1998:3), the SEM approach is “a statistical methodology
that takes a confirmatory (i.e. hypothesis-testing) approach to the multivariate analysis
of a structural theory bearing on some phenomenon”. A structural theory is used to
explain relationships among multiple variables or constructs. The processes in SEM
are represented by a series of structural equations and relations that can be modelled
pictorially to enable a clearer conceptualisation of the theory under study.
Schumacker and Lomax (2004:233) laid out a set of recommended steps for
data preparation before applying SEM, as follows:
• Establish a sound theoretical basis for the measurement models and structural
models in the study. This was set out in Chapter Four of this thesis.
• Clearly state the hypothesis for testing the structural model and/or alternative
models. This was also set out in Chapter Four of this thesis.
• Specify the type of correlation or variance-covariance matrix used in the
computer program (in this case, AMOS), and check for missing data, outliers,
non-normality and any issues that affect correlations. This will be done in
Chapter Seven.
• When using a correlation matrix, also include the means and standard
deviations of the observed variables to obtain the correct estimates of standard
errors for the parameter estimates. All those tests are reported in Chapter
Seven.
• Identify the estimation technique based on type of data matrix. This was
considered during the analysis.
Chapter Five – Research Methodology
-148-
Thus, through the SEM procedure, it is possible to achieve the simultaneous
examination and explanation of the pattern of a series of inter-related dependence
relationships among a set of latent (unobserved) constructs (Reisinger and Turner
1999).
5.10.4 Estimation Method
The Maximum Likelihood method was selected for each stage of the analysis.
It is the most common estimation method (Bollen 1989) in view of its reliance on
sample sizes in the 200–400 region (as opposed to the larger sample sizes required by
other estimation methods), and its reliability in situations with mild departures from
distribution normality (Gerbing and Anderson 1985).
5.10.4.1 Structured Models Comparison
There are several means of comparing structured models according to Steiger
et al. (1985) where researchers can assess the overall model fit by:
• Hypothesis testing using inferential statistical tests
• Likelihood ratio chi square test of exact fit – Compares target model to a
saturated (just identified model)
• Nested chi square tests for competing models
• Fit indices that indicate degree of fit
The two most common techniques used are the nested models and fit indices
that indicate degree of fit (Hair et al. 2008). According to Byrne (2002:128) nested
models are “...hierarchically related to one another in the sense that their parameter
sets are subsets of one another (i.e. particular parameters are freely estimated in one
model, but fixed to zero in a second model)”. The basic idea of comparison indices is
that the fit of a model of interest is compared to the fit of some baseline model. Even
though any model nested hierarchically under the target model (the model of interest)
may serve as a comparison model. The independent model is used most often because
it assumes that the observed variables are measured without error, i.e., all error
variances are fixed to zero and all factor loadings are fixed to one, and that all
variables are uncorrelated. However, nested models have their limitations because the
χ2 test is not only sensitive to sample size but also sensitive to the violation of the
Chapter Five – Research Methodology
-149-
multivariate normality assumption; hence, this study focused on fit indices because of
the large sample size (n = 529).
In support of the above, Steiger et al. (1985) noted that the majority of
methodological papers using SEM have focused mainly on using measures of fit
indices rather than on any other method (for example, nested models). They
concluded that the commonly used measures based on model comparisons are: the
Normed Fit Index (NFI ), the Non Normed Fit Index (NNFI ), the Comparative Fit
Index (CFI ), the Goodness-of- Fit Index (GFI ), and the Adjusted Goodness-of-Fit
Index (AGFI ), which this study has followed while applying SEM (see also, Jöreskog
and Sörbom, 1993).
5.10.5 Missing Data Treatment
Missing values can affect the input data matrix (the correlation or covariance
matrix) used by SEM. Missing data is usually dealt with through either imputation
(replacing missing data with a value chosen by the researcher), or pairwise or listwise
deletion (removing the cases for which some data is missing from some or all of the
analysis). Missing data were scrutinised to ensure they were not organised into any
particular pattern. In view of the small percentage of missing data (which ranged from
0 to 1.3%) and its apparent randomness, it was decided to choose imputation rather
than any deletion method, in order to preserve the sample size and avoid the
estimation problems often associated with the use of matrices of different sizes
engendered by pair-wise deletions.
5.10.6 SEM: Two-Step Method
In their pioneering article Structural Equation Modeling in Practice: A Review
and Recommended Two-Step Approach, Anderson and Gerbing (1988) proposed a
two-step model building approach. Their approach focused on the analysis of two
different models: the measurement model as a first step, followed by the structural
model. The measurement model specifies the relationships among the constructs and
their respective items. The structural model specifies relationships between the
constructs as described and identified by theory. Furthermore, the measurement model
tests discriminant and convergent validity while the structural model tests the
nomological validity of the research (Schumacker and Lomax 2004).
Chapter Five – Research Methodology
-150-
There have been several attempts in the literature to extend the two-step
approach to include further steps; for example, Mulaik and Millsap (2000) proposed a
four-step approach. Nevertheless, the majority of marketing related literature focuses
on one-step and two-step approaches, and, therefore, the research in this thesis only
examine these approaches as opposed to the other models.
While strictly speaking AMOS allows for both measurement and path models
to be tested (and re-specified) at once, many authors have advocated a two-step
approach (for example, Anderson and Gerbing 1988; Diamantopoulos and Winklhofer
2001; Schumacker and Lomax 2004). The first reason for this preference is a
compelling theoretical argument made by Jöreskog and Sörbom (1993:113): “the
testing of the structural model, i.e. the testing of the initially specified theory, may be
meaningless unless it is first established that the measurement model holds. If the
chosen indicators for a construct do not measure that construct, the specified theory
must be modified before it can be tested. Therefore, the measurement model should
be tested before the structural relationships are tested”.
The second reason stems from the manner in which AMOS models are
assessed. Since most of the relationships estimated in a AMOS model are those
leading from an observed variable to a latent variable (i.e. the measurement part of the
model) rather than the relationships linking latent variables (i.e. the structural part of
the model), the measurement part of the model plays a larger role than the structural
part in the overall fit of the model (Mulaik and Millsap 2000). Therefore,
misspecifications in the measurement model are best addressed and minimised before
the path model’s fit is estimated.
Even though existing literature suggests using the two-step method, this study
used a one-step approach as a consequence of the model fitting without the need to be
modified. This was a result of the initial development and validation of scales. A
degree of misspecification was expected; therefore, testing the whole model at once
would inevitably have led to poor fit indices, and the overall size of the model would
have made it difficult to trace the source of the main misspecifications.
Chapter Five – Research Methodology
-151-
5.11 Research Limitations
5.11.1 Procedure Error during Data Collection
Hair et al. (2006) pointed out a number of errors that may threaten the
reliability and validity of research (such as coding errors). These errors may lead the
researcher into drawing spurious conclusions. This section reviews the main types of
potential error, and the actions that can be taken to prevent them taking place. The
researcher’s potential source of bias is also examined.
5.11.2 Researcher Error
Any research can be affected by errors such as myopia (gathering the wrong
data because the problem has not been defined well), inappropriate analysis (omitting
relevant analysis, or conducting analysis in an inappropriate manner, or with data not
suitable for that type of analysis), misinterpretation (misunderstanding the
implications of the research, or being influenced by strong a priori ideas), or
communication (translating results in a manner which can be misunderstood, or
cannot be verified). Actions taken to prevent these potential errors include:
• An extensive review of the trustworthiness literature across various academic
disciplines. This resulted in the uncovering of a high level of convergence
between phenomena, supporting the structure of the conceptual model.
• The presentation of the conceptual model at several conferences (nationally
within the United Kingdom and internationally, for example, at ANZMAC), to
solicit reviewer and participant feedback and suggestions. This led to the
discovery of other relevant parts of the literature, and to the search for
additional support for the development of the hypotheses in Chapter Four.
• An extensive exploratory study, undertaken with the objective of investigating
all aspects of customer perceptions and understanding of trustworthiness.
• Providing exhaustive information and data about the collected results, and
detailing the steps taken to ensure the employed methods were used
appropriately. Furthermore, the thesis details all the results and tests carried
out in the analysis chapter, to enable readers to make their own assessments
regarding the final results.
Chapter Five – Research Methodology
-152-
5.11.3 Measurement Process
Measurement process errors may be due to conditioning (respondents may
behave differently from usual when their attention is drawn to a topic, and they may
over-elaborate), process bias (when respondents choose to answer questions in a
manner which is different from their true opinion), or errors in recording (when the
wrong answer is recorded).
Errors due to conditioning are the hardest to detect and correct or attenuate,
and to some extent all the social sciences are subject to this source of error. In
contrast, process bias and recording error situations are easier to detect. Steps taken to
guard against these errors included a consideration of individual cases of possible
process or recording bias. For example, if any case contained a clear indication from
the respondent of not wanting to answer all the questions and/or having answered the
questions in a clear systematic order, this case was identified as a case process bias
and excluded from the analysis.
5.11.4 Instrument Bias
Instrument bias is caused by individual questions (which can be ambiguous,
difficult to understand or confusing) or by the instrument and in particular the order of
the items. To minimise the sources of instrument bias in individual items, the
following measures were adopted. Items were developed using guidelines provided in
the literature (DeVellis 1991) about the wording of items:
• The wording of certain items from existing scales was adjusted to make sure
the wording of the questionnaire was jargon-free, so the respondents could
understand the questions easily.
• Items written for scales developed in this study were submitted to three expert
judges10
10 See appendix Ten for the judges description.
, and one of the tasks required of them was to suggest which items
should be omitted if they were unclear or had other flaws that would limit their
validity.
Chapter Five – Research Methodology
-153-
• The very thorough scale development and validation process also screened out
further confusing items, for example, during the EFA and reliability test
stages.
Conscious of the possibility that respondents can be ‘led’ to answer in a
specific way when items related to specific scales are presented in the same order as
the sequencing of the hypothesised relationships, items from all scales were presented
in a random order, with no two items from the same scale presented one after another.
5.11.5 Respondent Error
Respondent error may be due to response styles, which are “tendencies to
respond systematically to questionnaire items on some basis other than what the items
were specifically designed to measure” (Grix 2004:120) or response error (which may
include uncertainty when respondents are unsure about their real opinion, in
articulation, or mistakes). Response styles can contaminate respondents’ answers, and
in the process jeopardize the validity of conclusions drawn about the validity of new
measures or the relationships between measures.
Summary
Since the current research is using a mixed method, applying survey methods
(interview and questionnaire) in both its parts appears to make it easier to rationalise
the research, because any researcher who wants to view a phenomenon in its broadest
sense, with a large population, would achieve significant benefit from the use of
surveys.
Concerning the scale procedure, this thesis adopted the C-OAR-SE scale
development procedure since it appeared to be applicable to the rationale of the
research, keeping in mind different criticisms of the scale and the fact that the scale
has not been empirically tested and applied in different disciplines. Furthermore, in
view of this research applying the C-OAR-SE procedure, content validity will be
examined over the other forms of validity.
Chapter Six – Scale Development and Pilot Study
154
CHAPTER SIX: SCALE DEVELOPMENT
AND PILOT STUTDY
6.1 Introduction
Empirical social scientists focus on measuring attributes of objects that are
abstract in their nature (Netemeyer et al. 2002). Such abstract objects and their
associated attributes cannot be observed or quantified directly, and are often
considered to be latent. Latent constructs can also be variable, i.e. their attributes may
change over time; for example, changes in customer satisfaction over a period of time.
Therefore, the measurement of latent constructs requires the researcher to develop a
scale to estimate the actual value at a given point in time (Netemeyer et al. 2002). It is
generally agreed that latent constructs require multiple items or statements to fully
capture the meaning of the latent variable (DeVellis 2004).
This chapter reports the different stages in the development and validation of
the measurement scale for this thesis used to test the main hypotheses established in
Chapter Four. Initially, a pool of ninety items was generated from the literature
describing previous research on trustworthiness and its proposed antecedents. Next a
pilot study was carried out to purify the pool of items and to assess the reliability of
the remaining items. The pilot consisted of three stages, namely; interviews with a
panel of judges; semi-structured interviews with customers; and a structured
questionnaire to test the reliability of the research scale. The results of the three stages
were positive, which affected the research direction and provided a solid base on
which to build the next stage of the study.
Chapter Six – Scale Development and Pilot Study
155
6.2 Methods of Enquiry
Research methodology attempts to achieve an “examination of the possible
plans to be carried out, the journeys to be undertaken (in research), so that
understanding can be obtained” (Polkinghorne 1983:5). These methodologies include
aspects such as sampling principles, statistical analysis and, most importantly, issues
surrounding the validity and reliability of the research.
It is accepted that the methodological procedures that a researcher might
employ are significantly influenced by the assumptions, interests, values and goals of
the researcher. Positivists and interpretivists, for instance, hold different assumptions,
interests, values and goals, and therefore, they would apply different methodologies
due to their different approaches to problems and their search for different answers
(Creswell 1994).
A methodology that has been developed for use in this thesis is a multi-method
triangulated approach towards data collection. This approach entails several stages to
ensure the appropriateness of the data collection, which in turn helps to enhance the
validity and reliability of the research. These stages are demonstrated in Figure 6.1.
Define construct
Define the antecedents of trustworthiness
Interviews with a panel of 3 judges
Refine the pool of items
Develop and undertake pilot questionnaire
Analysis from interviews
11 in-depth interviews
and card sort
Face-to-face self-completion questionnaire
60 respondents
Design final questionnaire
Report findings and conclusions
Figure 6.1: Overview of the Scale Development Process and the Pilot Study
Chapter Six – Scale Development and Pilot Study
156
6.3 Scale Development
The important parameters of this thesis have already been outlined, but any
attempt to test the proposed framework1
To avoid these limitations, this thesis embraces content validity as a main
validity indicator, and follows Rossiter’s (2002) approach to scale development.
However, the thesis attempts to achieve a balance between the two approaches by
reporting some tests from Churchill’s (1979) procedure that were thought suitable to
add greater reliability and validity to the research instrument, for example, EFA
analysis. However, the C-OAR-SE scale development method advocated by Rossiter
(2002) was the main procedure used for scale development
requires a set of measurement instruments for
the key construct and formed attributes. Rossiter (2002) questioned Churchill’s (1979)
traditional approach to the development of scales for measuring marketing constructs,
based on a multi-trait multi-method approach, and instead proposed a six-step
procedure called C-OAR-SE. Churchill’s (1979) procedure allowed the internal
reliability of a scale to be measured in an objective manner by using Cronbach’s alpha
coefficient, but this test of reliability does not necessarily imply that the scale is valid.
As Rossiter (2002) argues this shortcoming throws into doubt the validity of some
well-known scales, for example SERVQUAL, with its emphasis on exploratory factor
analysis. The C-OAR-SE approach focuses on the conceptualisation of constructs,
thereby addressing a weakness in Churchill’s (1979) procedure. In their work, Finn
and Kayande (2005) critically examine The C-OAR-SE; they support the attention to
conceptualisation and content validity, but criticise the extreme context dependence,
which is a point also raised by Diamantopoulos (2005).
2
According to Rossiter’s (2002) C-OAR-SE procedure, the traditional scale
development process normally involves the generation of a number of potential scale
items and the completion of a confirmatory factor analysis to identify which items to
include or exclude. Rossiter (2002) puts forward a compelling argument to indicate
in this thesis, because it
relies on content validity to support the proposed constructs over being data driven as
been advocated by Churchill (1979).
1 See Chapter Four - Section 4.6. 2 See Chapter Five - Section 5.7 for a full discussion on the C-OAR-SE scale development procedure.
Chapter Six – Scale Development and Pilot Study
157
that this approach is inconsistent because it can lead to the inclusion of scale items
that have little relevance to the construct. The key to the validity of a scale is its
content validity, and therefore, an open-ended approach is suggested to identify which
items to include.
Table 6.1 shows some of the empirical studies on trust and trustworthiness. A
detailed view of these studies is reported, including information about the
measurement instrument, the research context and method of analysis. This thesis
follows in the footsteps of these studies, applying a Likert scale, using Cronbach’s
alpha for reliability assessment, and employing SEM for the main analysis.
In line with Rossiter’s (2002) suggestion, the first stage of the scale
development procedure involved interviews with a panel of three judges/experts3 to
assess the scale items that were generated from the existing literature. The second
stage involved eleven individual one-to-one in-depth interviews4 concerning the
nature of trustworthiness for service organisations5
Table 6.1 lists some key empirical studies in the field of marketing. The table
reports how these studies were conducted; more specifically, it reports the items that
were accepted as being relevant for the construct of trustworthiness. The pilot
questionnaire uses an interval scale to assess the construct, its attributes and
determinants. This table is important because it clearly shows the interests of the key
researchers in marketing and how they carried out their research; this gave clear
. This led to the next phase that
involved each respondent taking part in a card sort exercise. To develop the scale
items for the card sort, an initial pool of ninety items (drawn from qualitative work
and existing studies) was subjected to screening techniques to create a revised pool of
thirty-six items. Each member was asked to sort the items into piles according to how
they felt each item related to specific determinants of trustworthiness in the proposed
model, as well as the attribute trustworthiness and the construct of trust. The results of
the card exercise produced a distribution of items that was broadly consistent with the
thesis assumptions. These findings were taken to indicate a good degree of content
validity.
3 See appendix Ten. 4 The participant selection procedure for this second stage consisted of randomly selecting customers from two hotels in Jordan. 5 More details in Section 6.4.1.
Chapter Six – Scale Development and Pilot Study
158
direction in the development of this thesis on how to conduct the research, which
scale to adopt and which statistical tests to use.
Chapter Six – Scale Development and Pilot Study
159
Table 6.1: Main Empirical Studies on Trust and the Scales Used Author/s Research context Dimensionality
of the construct Number of items
Employed measurement scales
Data analysis methods and principal results
Anderson and Weitz
(1989)
Data collection from 95 sales agencies with reference to their relationships
with690 manufacturers in the electronic components sector (from the
perspective of the sales agencies)
Trust 2 items 7-point scale: 1 = very little trust;
7 = great deal of trust
System of equations estimated via three stage least square). Cronbach alpha = 0.84. The correlation coefficients between trust, on one hand, and perceived competency
and goal congruence, on the other, are 0.51 and 0.69 respectively.
Anderson and Narus
(1990)
Data collection from 249 distributors and 213 manufacturers (from the
perspective of both firms)
Trust (on the distributors’ side) 3 multi-
item indicators (on the
manufacturers’ side) 4 multi-
item indicators
7-point scale: 1 = don’t trust X;
7 = trust X completely
Structural equations models – Lisrel. In both samples, both measurement and structural model required several re-
specifications
Morgan and Hunt (1994)
Study of 204 relationships between automobile tyre retailers and their
suppliers (from the perspective of the buyers)
Reliability integrity
7 items 7-point scale: 1 = strongly disagree
7 = strongly agree
Structural equations models – Lisrel. Cronbach alpha = 0.95 The correlation
coefficient between trust and opportunistic behaviour is -0.759
Mohr and Spekman (1994)
Study of 102 vertical relationships between personal computer retailers
and their suppliers (from the perspective of the retailers)
Trust 3 items 5-point scale: 1 = strongly disagree
5 = strongly agree
Multiple regression. Cronbach alpha = 0.75.
Nielson (1998)
Study of 163 relationships between manufacturers and distributors of
intermediate products such as components parts, raw materials etc. (from the perspective of the seller)
Trust 3 items 5-point Likert scale: 1 = strongly disagree
5 = strongly agree
Structural equations models – Lisrel. Cronbach alpha = 0.87 The measurement
scale of trust is judged to be acceptable for testing the causal model
Selnes (1998)
Study of 177 relationships between restaurants and their suppliers (from
the perspective of the buyer)
Trust 1 item Scale: 1–10 1 = strongly disagree 10 = strongly agree
Structural equations models – Lisrel. Correlation between competence and trust is 0.464. Competence does not have any
effect on trust (it acts through communication, highly correlated)
Chapter Six – Scale Development and Pilot Study
160
Crosby Evans and Cowles
(1990)
Study of 151 relationships between insurance policies buyers and sales agents (from the perspective of the
buyer)
Trust 9 items 7-point scale: 1 = strongly disagree 7 = strongly agree
Structural equations models – Lisrel. Cronbach alpha = 0.89 CFA validated the
hypothesised factor structure
Moorman et al. (1992)
Moorman et al. (1993)
Study of 779 relationships between providers and users of market research
(from the perspective of the users)
Belief behavioural
intention
5 items 5-point scale: 1 = strongly disagree
5 = strongly agree
Regression analysis, with results then used as inputs for a path analysis. When
subjected to factor analysis, all items of trust loaded highly on a single factor,
supporting the uni-dimensionality of the construct. Cronbach alpha = 0.84
Doney and Cannon (1997)
Study of the relationships between a sample of 210 industrial firms with supplier firms and their salespeople (from the perspective of the buyer)
Credibility benevolence
Trust towards the firm: 8 items; trust towards the
salespeople: 7 items
7-point scale: 1 = strongly disagree;
7 = strongly agree
CFA with Lisrel and system of equations estimated with three-stage least squares. Cronbach alpha towards supplier firm =
0.94, towards the salespeople = 0.90. The measures used show good psychometric
properties. Trust towards the firm and trust toward salespeople are distinct constructs.
Sirdeshmukh et al. (2002)
Two service organisation, retail clothing (n= 264) and non-business
airline travel (n = 113)
Trustworthiness 4 items 10-point scale Semantic Scale
Path analysis using EGS software reporting Cronbach alpha for reliability
testing
Chapter Six – Scale Development and Pilot Study
161
6.4 Overview of the Pilot Study
A pilot study is often referred to as a small-scale research project that collects
data from respondents similar to those which will be used in the full data collection
(Zikmund and Babin 2007:62). This type of study is a critical technique in refining
research measures and reducing the risk that the full study will be fatally flawed.
Hence, the pilot study is seen as an early indicator of potential effectiveness of the
research by developing a mini sample and attempting to test the research objectives.
The pilot study for this thesis aimed to test:
• How respondents interpreted the questionnaire.
• The level of their understanding of the questions and whether there was any
misleading wording (Creswell 1994).
• The reliability of the measurement scale by applying Cronbach’s coefficient
alpha (α)6
As mentioned earlier, the pilot study incorporated four distinct but interrelated
data collection phases. Using a mixed method instrument, interviews with a panel of
three judges took place in order to purify the generated pool of items (n=90) and to
comply with the research procedure advocated by Rossiter (2002) (see Table 6.2 for a
full overview of the data analysis in the pilot study).
.
During the first stage, semi structured interviews were carried out
independently with three judges to review the list of the generated items in order to
eliminate similar items from the original pool. The panel of three judges consisted of
two academics and one industry expert in marketing. The selection of judges7
• Indicate which items they felt should be omitted because of some flaw (and
they were invited to provide further information/suggestions about these
items).
was
based on availability and expertise; each judge was free to alter, delete or add any
item to the list. They were required to fulfil two main tasks:
6 See the discussion in Chapter Five - Section 5.9.2. 7 See appendix Ten for the judges position and country description.
Chapter Six – Scale Development and Pilot Study
162
• Sort the items based on the definition of each construct. The items were
supplied to them in a random sequence.
A number of items were recorded for the pilot, based on expert feedback and
the importance of including negatively-worded items in each scale. Following this, the
panel of judges were required to evaluate the relationship between the research items
and the proposed constructs in order to assess the pertinence of the items. It is worth
noting that none of the changes were adopted unless at least one more judge agreed on
the alteration of the item(s). The results were significant in that several items were
deleted (delete items = 49, new pool = 41) and adapted because they were thought to
be too similar, too repetitive, or not relevant to their respective construct.
Table 6.2: Detailed view of the Data Analysis Strategy
Activity Analytical technique(s) used Purpose Identify the research constructs
Literature review Conceptualisation based on extensive exploratory study
To clearly identify the construct of trust in the context of the surrounding theory, to determine the relevant theory for this research
Generate a pool of items Review of existing scales Review of exploratory study Item editing
To generate a large pool of items for each construct (n=90)
Expert judging (3 judges)
Sorting task (experts asked to allocate items to the construct they believed they are related to the relative construct)
To refine the pool of items in the light of the judges’ views
Questionnaire pilot (60 respondents)
Scale reliability Questionnaire clarity Exploratory Factor Analysis
To examine the extent of the reliability of the scale, along with detecting any vagueness in the questionnaire
Collected data (n = 556) Measure purification – internal consistency
Exploratory Factor Analysis Construct Reliability
To obtain, for each measure, a set of items which are internally consistent
During stage two, interviews were carried out with randomly selected
customers who had experience with the service provider (n = 11). Card sort exercises
were carried out to ask customers about the importance of the generated pool of items
(n = 41) in terms of to what degree they see the items significant and applicable to the
Chapter Six – Scale Development and Pilot Study
163
related construct. Card sorting is a technique used for exploring how individuals
group and prioritise items, so that the researcher can develop structures that maximise
the probability of customers being able to find and rank items, and can also identify
whether customers understand the terminology of the research. Card sorting is an easy
and cheap method to employ, and it enables the researcher to understand how ‘real
people’ are likely to group items, identify items that are likely to be difficult to
categorise, and find terminology that is likely to be misunderstood.
The collected data demonstrated a clear indication of the relevant items that
measure the components of trust. This identification led to stage four of the pilot
study, where a questionnaire was developed and administered based on the previous
three stages to assess the internal consistency of the scale (n = 60). The final results of
the pilot study indicated general approval of the projected scale and its reliability, and
resulted a pool of the items (n = 36) that measure the proposed constructs of the study.
6.4.1 The Interviews
Interviewing was selected as a technique for the pilot study because it “can
provide a greater depth of data than the other types, given its qualitative nature”
(Denzin and Lincoln 2000:652). A semi-structured format was chosen to incorporate
the formulation of open questions8 based upon the initial conceptual framework9
and
the review of the literature, while at the same time encouraging participants to provide
a greater depth of information on issues that they perceived to be of relevance.
The semi-structured sections of the interviews were intended to gather
information on themes derived from the initial conceptual framework in the first stage
of the pilot study, while the second part of each interview consisted of card sort
exercises where the participants were asked to rank items according to their
importance and match them with associated constructs. However, both stages aimed
to assemble information on a range of topics drawn from the literature (see Table 6.3).
8 Open questions were employed to allow respondents to freely express their views on the study. 9 See Chapter Four – Section 4.6.
Chapter Six – Scale Development and Pilot Study
164
Table 6.3: Examples of the Literature Supporting the Research Constructs10
Construct
Supporting literature Integrity
De Wulf et al. (2001) Hess (1995), Morgan and Hunt (1994) Cummings and Bromiley (1996)
Benevolence
Bhattacherjee (2002) Hess (1995), Morgan and Hunt (1994) Bearden and Netemeyer (1999)
Competence
Cummings and Bromiley (1996) Jarvenpaa et al. (2000) Sirdeshmukh et al. (2002)
Value alignment
Siegtist et al. 2002 Hess (1995)
Consistency
Anderson and Narus (1990) Cummings and Bromiley (1996) Garbarino and Johanson (1999)
Communication
Anderson and Narus (1990) Morgan and Hunt (1994)
Trustworthiness
Bhattacherjee (2002) Cummings and Bromiley (1996)
Behavioural loyalty Rundle-Thiele and Bennett (2001) Dick and Basu (1994)
Attitudinal Loyalty Dick and Basu (1994) Rundle-Thiele and Bennett (2001)
The aim of the interviews was to reduce the initial set of items that were
generated from the literature. No new items were generated for the current research
based on these interviews because the set of items generated by an examination of the
literature was adequate to cover the proposed variables of the study, remembering that
in order to analyse each construct, SEM requires a minimum of three items for that
construct. Since the literature generated a sufficient number and range of items, the
generation and testing of further items did seem to be necessary.
The groups of participants for the interview stage of the pilot study were
randomly selected at the time of the study, and initial contact at each stage was made
by the use of simple random sampling in that each selection from the population was
“entirely independent of the next” (Cohen et al. 2001:100). Selected hotels were
contacted in Jordan and arranged a time for the interviews with the hotel customers
inside the hotel, to minimise disruption for both the participants and the hotel. The
selected participants were approached in the hotel lobby and were introduced to the 10 This information is also available in an expanded form in Chapter Four - Section 4.6.
Chapter Six – Scale Development and Pilot Study
165
research idea. Once the participants had agreed to take part in the interview; the
participant and a representative from the hotel proceeded to a meeting room in the
hotel to begin the interview.
All eleven interviews were conducted individually, and a rapport was
established with each participant so that they would feel comfortable sharing their
thoughts and feelings. All the definitions of key terms were clarified, and an
explanation of how participants’ comments would contribute to the overall research
objectives. Initially some time was spent talking informally about background and
interests, and then the card sort exercises were introduced, using cards and a white
board pre-designed by the researcher. It was found that sharing the researcher’s
background of professional practice within the hotel sector created a collegial
atmosphere, and participants seemed to be drawn to divulge greater detail knowing
that the researcher had an intimate understanding of the industry. A decision was
made not to make audio recordings of the interviews since it was thought that
participants might be less likely to be open and truthful if recorded. Instead, extensive
interview notes were taken, and were then transcribed immediately following each
interview.
Prior to the interview trials, the interview schedule was pre-tested with three
respondents11
Originally a list of ninety items was derived from previous studies surrounding
trust and trustworthiness constructs. These items were linked to the constructs they
were thought to refer to. The panel of three expert judges were interviewed
. They read the schedule and provided feedback on the intention,
wording and sequencing of the instrument. Feedback on the schedule was considered
and changes were made accordingly; for example, the respondents made some
comments in relation to the terminology used, as a result of which the definitions of
key terms were given and explained to each participant at the beginning of their
interview. The language that had initially been used was altered to suit the participants
and eliminate any possible misunderstanding of the questions.
11 The three respondents were two friends and one family member.
Chapter Six – Scale Development and Pilot Study
166
individually to assess both the similarity of these items and the relationship between
the items and their associated constructs12
Participants responded to questions assessing their knowledge of the service
provider, its values, staff, and any other issues related to the perceived trustworthiness
of the service provider. Their responses were recorded on a 5-point Likert scale
anchored using strongly agree and disagree. The mid-point of the scale was anchored
using the word neutral, but there was no wording for points 2 and 4 of the scale.
.
Overall, the results of the expert interviews purified the list to forty-one items
related to the construct of the study. This is still considered a large number of items
according to Hair et al. (2006). As an example of a deleted item, most participants
(n=11) did not believe that the item ‘my ….. is truthful’ was important from their
perspective, and it was ranked lowest in importance in the integrity construct, despite
literature approval by Hess (1995) and Morgan and Hunt (1994). Therefore, this item
was deleted in the context of the service sector; it’s not advocated in this thesis that
this item is not effective in measuring integrity, but simply that it is not applicable to
this study or the hotel sector.
The results of these expert interviews were carried forward to the next stage of
the research involving the semi-structured interviews with customers from the hotel
sector carrying out a card sorting exercise. The participants were asked to rank each
item according to its importance from their perspective. The results from the semi-
structured interview stage purified the pool of items to 36 items. In the next stage each
construct was represented only by the four most relevant items.
6.4.2 The Questionnaire
The third stage of the pilot study contained a structured questionnaire
administered to a randomly selected sample of customers in the hotel sector. The idea
behind structuring a questionnaire grew from the fact that questions in such an
instrument may be designed to collect either qualitative or quantitative data. Any
question must be clearly described, but in particular questions that assess a qualitative
measure must be carefully phrased to avoid uncertainty.
12 See the earlier discussion in section 6.4 for more details.
Chapter Six – Scale Development and Pilot Study
167
Rossiter’s (2002) criticisms of Likert scales were taken into consideration, but
piloting suggested that respondents were comfortable with this format and interpreted
the anchors appropriately. Initially, customers responded to questions assessing the
degree to which certain factors were important in influencing their decision to indicate
the service provider’s integrity; these factors included; show fairness in transactions;
consistency in service delivery; keeping promises; and presenting respect. The rest of
the questions assessed the degree to which respondents perceived their service
provider to display consistency, competence, benevolence, value alignment and trust;
these questions documented the influence of experience, current service, autonomy,
comfort, and personal values.
The second part of the questionnaire was designed to determine what, if any,
external factors would influence a respondent to place their trust with the service
provider. The questions assessed safety, brand, location, and atmosphere of the
location of the service; these factors were not included in the trust literature, but they
can be seen as an indication of customers’ loyalty towards the service provider.
The last section simply asked the respondents for some demographical
information, which could be correlated with their views in order to identify in-depth
relations; these questions were about age, gender and nationality. For the purposes of
this study, it was decided not to ask further demographical questions, because the aim
of the research was to confirm the proposed model rather than to explore the service
sector.
Self-completion questionnaires were employed in the pilot13. A sample of five
“dummy”14 questionnaires was distributed15
13 The questionnaire was administered in English with an Arabic version available upon request, however, all participants used the English version in both the pilot and the main study, 14 It is common practice to use “dummy” questionnaires prior to collecting the main data in order to avoid errors and misinterpretation of the questions (Creswell 2003). 15 The five respondents consisted of five friends.
to identify any problems in interpreting
the questionnaire instructions or items. Since multivariate techniques are sensitive to
sample size effects, larger sample sizes increase confidence in assessments of
construct validity and reliability. Several item-to-sample ratios have been proposed,
but as long as there is sufficient intercorrelation, a smaller sample size can be used
(Hair et al. 2006). A total of sixty responses were collected, just above the cut-off
Chapter Six – Scale Development and Pilot Study
168
point for factor analysis suggested by Hair et al. (2006). The sixty completed
questionnaires were distributed to randomly selected hotel customers in two five-star
Jordanian hotels.
In spite of Rossiter’s (2002) argument against the use of formal measures of
reliability when developing scales to test formed constructs and attributes, the result
of the reliability analysis is reported here as a further check on the quality of the
proposed measurement structure. The reliability analysis test showed that the overall
Cronbach alpha coefficient for the scale (α) is .79. In more detail, the reliability
analyses showed that the Cronbach alpha coefficient (α) for the items related to
consistency was .80, competence .78, integrity .91, benevolence .90, value alignment
.76, communication .85, trustworthiness .75, behavioural loyalty .82 and attitudinal
loyalty .76. All the reliability coefficients are considered to be acceptable (Hair et al.
1998).
Based on the stated objectives and the achieved results of the pilot study, the
pilot stage was successful. This study paves the way for the main data collection and
analysis stage of the research.
During the pilot study it was discovered that some respondents misinterpreted
some questions in the questionnaire even though an Arabic explanation was offered.
Therefore, immediate corrections were carried out to reword these questions to suit
the respondents. In addition, it became apparent to the researcher that the
questionnaire did not include a question on the hotel grade, which might have been
useful in the main survey.
Chapter Six – Scale Development and Pilot Study
169
6.5 Scale Evaluation
The previous stages of scale development were devoted purely to the
development of a new scale. DeVellis (2004) describes validation guidelines for
scales, and make a distinction between techniques for: a) content validity; b) construct
validity; and c) reliability. The C-OAR-SE procedure does not explicitly mention the
need for an extra sample, but it does place emphasis on the content validity of the
items. Churchill’s (1979) traditional procedure, in contrast, specifically advocates a
second data set to assess reliability, i.e. the stability of the instrument across different
measurements and preferably also across different methods. Construct validity should
also be re-evaluated; CFA can be used to test discriminant and convergence validity.
In addition to the above, this thesis followed Hinkin’s (1995)
recommendations by using CFA to test criterion-related validity (predictive or
nomological validity). This was done by confirming some theoretically hypothesised
relationships using SEM. Before the scale was developed a careful reanalysis of
content validity was deemed to be necessary. This data analysis resulted in the
omission of nearly half of the items from the scale (forty out of ninety). The next step,
according to the C-OAR-SE procedure, is to assess content validity, which has been
achieved in defining each of the proposed constructs16
. Furthermore, psychometric
tests will be reported in the next two chapters. For each of the nine constructs, an
evaluation was made concerning whether the remaining items are sufficiently
representative of their respective constructs. The remaining items used in the main
survey are reported in Table 6.4.
16 See Chapter Four - Section 4.6.1.
Chapter Six – Scale Development and Pilot Study
170
Table 6.4: The Final Research Items Item Determinant Reference/Source Show fairness in transactions Integrity Bhattacherjee (2002) Always achieves consistency in service delivery Integrity New item Always keeps its word Integrity Cummings and Bromiley
(1996) The ….. employees treat me with respect Integrity Sirdeshmukh et al. (2002) Is open to my needs Benevolence Bearden and Netemeyer
(1999) Acts in a caring manner Benevolence Hess (1995) Keeps its customers’ best interest in mind during most transactions
Benevolence Bhattacherjee (2002)
Is receptive to my needs Benevolence Bhattacherjee (2002) Has adequate skills to deliver the right service Competence Jarvenpaa et al. (2000) Is efficient Competence Sirdeshmukh et al. (2002) The ….. employees can competently handle most of my requests
Competence New item
The ….. employees can be relied upon to know what they are doing
Competence Sirdeshmukh et al. (2002)
Shares the same values as me Value alignment Siegrist et al. (2002) Is ethical while dealing with me Value alignment Morgan and Hunt (1994) Is interested in more than just making profit out of customers
Value alignment Hess (1995)
The employees act as if they value the customer Value alignment New item Has stability in its operations Consistency New item Is always reliable Consistency Hess (1995) Always matches customers’ expectations Consistency Garbarino and Johanson
(1999) Always satisfies customers’ needs Consistency New item Communicates clearly Communication Ennew and Sekhon (2005) The ….. respond immediately when contacted Communication New item Inform me immediately of any problems Communication Anderson and Narus (1990) The ….. always communicates with me Communication New item I will transact with this ….. again for future visits Behavioural
loyalty Ruyter et al. (1998)
I will try new services that are provided by this ….. Behavioural loyalty
Rundle-Thiele and Mackay (2001)
I will recommend other people to visit this ….. Behavioural loyalty
Rundle-Thiele and Mackay (2001)
I will say positive things to other people about the services provided at this …..
Behavioural loyalty
Rundle-Thiele and Mackay (2001)
I will continue to visit this …… even if the service charges are increased moderately
Attitudinal loyalty
Ruyter et al. (1998)
I have strong loyalty to this ….. Attitudinal loyalty
New item
I will keep visiting this ….. regardless of everything being changed somewhat
Attitudinal loyalty
Ruyter et al. (1998)
I am likely to pay a little bit more for using the services of this …..
Attitudinal loyalty
Ruyter et al. (1998)
The ….. employees are professional and dedicated to customers
Trustworthiness New item
The ….. respond caringly when I share my problems
Trustworthiness New item
My ….. is always honest with me Trustworthiness Morgan and Hunt (1994) Overall I feel I can trust my ….. Trustworthiness Ennew and Sekhon (2005)
Chapter Six – Scale Development and Pilot Study
171
Table 6.4 reports the final scale items after the purification process. Four items
were selected to measure each of the proposed constructs which is the recommended
number as suggested by (Hair et al. 2006), the majority of these items are
theoretically supported, with the exception of eight new items (see Table 6.4) that
were added in response to the judges and the in-depth interviews. A full discussion on
each construct with its respective items follows below.
1. Integrity
The content validity of the items employed to measure construct integrity is
acceptable. The original conceptualisation of integrity is contained by the items that
load highly on the first factor. Four items were dropped during the interview process –
for example, ‘my .… will not deceive’ and ‘my …. is truthful’. Three items were
dropped after the judges’ feedback – for instance, ‘my …. always encompasses
consistency in service delivery’. The EFA indicated that the remaining items are part
of the same underlying construct17
2. Benevolence
.
The content validity of the items used to measure the construct benevolence
was achieved. The original conceptualisation of benevolence is still covered by the
items that load highly on the first factor. Six items were dropped in the interview
process – for example, ‘my …. acts in a caring manner’ and ‘my …. will act in
goodwill’. Four items were dropped after the judges’ feedback – for instance, ‘my ….
always has empathy’ and ‘my …. has policies that favour the customer’s best
interest’. The EFA indicated that the remaining items are part of the same underlying
construct.
3. Competence
The content validity of the items used to measure the construct competence is
satisfactory. The original conceptualisation of competence is still covered by the items
that load highly on the first factor. Six items were dropped in the interview process –
for example, ‘my …. has the right knowledge’. Three items were dropped after the
17 The full EFA report is available in Chapter Eight – Section 8.3.1.
Chapter Six – Scale Development and Pilot Study
172
judges’ feedback – for instance, ‘my …. does not give unrealistic promises’ (reverse
item), ‘my …. is able to meet customer needs’ and ‘my …. is efficient’. The EFA
indicated that the remaining items are part of the same underlying construct.
4. Value alignment
The content validity of the items employed to measure the construct value
alignment is satisfactory. The original conceptualisation of value alignment is still
covered by the items that load highly on the first factor. Five items were deleted based
on the judges’ recommendation – for instance, ‘my …. shares similar beliefs with
customers’. The EFA indicated that the remaining items are part of the same
underlying construct.
5. Consistency
The content validity of the items used to measure the construct consistency has
met the requirements for defining this construct. The original conceptualisation of
consistency is still contained by the items that load highly on the first factor. Nine
items were dropped in the interview process – for instance, ‘for my …. there are no
limits to how far they will go to resolve customers’ problems’. The EFA indicated that
the remaining items are part of the same underlying construct.
6. Communication
The content validity of the items applied to measure the construct
communication is acceptable. The original conceptualisation of communication is still
covered by the items that load highly on the first factor. Thirteen items were agreed to
be dropped by the judges as they did not relate to the construct of communication;
therefore, they were all dropped. This may relate to the number of available published
scales that measure communication. For example, the statement ‘my….
communicates steadily’ was dropped. The EFA indicated that the remaining items are
part of the same underlying construct.
Chapter Six – Scale Development and Pilot Study
173
7. Behavioural loyalty
The content validity of the items employed to measure the construct
behavioural loyalty was achieved. The original conceptualisation of behavioural
loyalty is covered by the items that load highly on the first factor. No items were
deleted due to general acceptance of their importance to the construct from all the
interviewees and judges. The EFA further confirmed this and indicated that the items
are part of the same underlying construct.
8. Attitudinal loyalty
The content validity of the items used to measure the construct attitudinal
loyalty is acceptable. The original conceptualisation of attitudinal loyalty is still
contained by the items that load highly on the first factor. No items were deleted due
to general acceptance of their importance to the construct from all the interviewees
and judges. The EFA further confirmed this and indicated that the items are part of the
same underlying construct.
9. Trustworthiness
The content validity of the items employed to measure the construct
trustworthiness is adequate. The original conceptualisation of trustworthiness is still
covered by the items that load highly on the first factor. No items were deleted due to
general acceptance of their importance to the construct from all the interviewees and
judges. The EFA further confirmed this and indicated that the items are part of the
same underlying construct.
The final table of items forms the basis of the main questionnaire for this
thesis. The questionnaire was discussed in 6.4.2. The full results from the data
collection will be reported and discussed in detail in the next two chapters.
Chapter Six – Scale Development and Pilot Study
174
Summary
A pilot study with three stages was carried out in order to validate the research
and resolve some of the problems that may occur in the main survey. Stages one and
two consisted of interviews, firstly with a panel of three judges who hold vast
experience in the industry and who raised a number of essential points for the
researcher to consider, and secondly with a randomly selected sample of customers in
the service sector in order to gain an in-depth insight into their views on
trustworthiness. The last stage comprised a survey questionnaire to assess the
reliability of the measurement scale.
The pilot study found that there was consistency between the proposed
constructs and their associated items. The pool of items started with ninety different
statements to measure the different dimensions surrounding trust; these items were
reduced to sixty after passing the panel of judges, the rejected items being not relevant
or showing significant similarity with others. The pool of items was further reduced to
contain only the four strongest items related to each variable; this proved to be the
most adequate method in order to apply SEM. The scale reliability analysis showed a
satisfactory result by scoring .79.
The adopted approach in the pilot study has several drawbacks; for example,
the pilot study conducted a card sorting excise with customers without any in depth
interviews to allow customers to express their views freely. Moreover, the pilot study
was limited in the sample size (n = 60) which did not allow the application of further
tests to EFA; for example multiple regression analysis.
The next chapter reports the outcome of the final questionnaire. It will start by
exploring the different descriptive results, as well as applying confirmatory factor
analysis and SEM.
Chapter Seven – Descriptive Analysis of Measurement Scale Items
-175-
CHAPTER SEVEN: DATA ANALYSIS AND
RESULTS – DESCRIPTIVE ANALYSIS OF
MEASUREMENT SCALE ITEMS
7.1 Introduction
Building on the qualitative data from the in-depth interviews and the results
from the pilot study discussed earlier1, the purpose of this analysis is to generalise the
results from the main survey and confirm the proposed model2
This research used a self-completion survey technique established by the
results of the pilot study
. The main reasons for
selecting a survey questionnaire are provided by Cramer (2003), who noted that
carrying out a survey questionnaire holds distinctive advantages over other data
collection techniques (for instance interviews), such as: “economy of the design, the
rapid turnaround in data collection, and the ability to identify attributes of a
population from a small group of individuals” (Creswell 1994:119).
3
1 See Chapter Six for the pilot study and scale development. 2 The model is listed and discussed in Chapter Four section 4.6.1 and illustrated in Figure 4.8. 3 Contained in Chapter Six - Section 6.4.
. Once the final measurement scale and the survey
questionnaire had been developed, the survey was sent to the selected hotels in
Jordan. When the responses had been gathered a detailed analysis was conducted to
provide a thorough understanding of the results and the significant outcomes of the
model.
The data analysis for this thesis is divided into two main parts. The first part is
contained in this descriptive chapter, and is mainly concerned with the characteristics
of the collected sample. The second part consists of the application of a Structural
Equation Modelling (SEM) procedure forms the empirical examination of the
proposed model and presented in Chapter Eight.
Chapter Seven – Descriptive Analysis of Measurement Scale Items
-176-
In this chapter the collected data from the main quantitative survey will be
presented. The chapter begins with the steps of designing, developing and
administering the questionnaire, and then it describes the data screening process that
took place prior to statistical analysis. Issues related to missing data and outliers will
be addressed.
7.2 Data Collection Sample and Response Rate
According to the Jordanian Government statistics4, the number of tourists that
used hotels in Jordan between April 2006 and April 2007 was approximately 1.1
million5
The targeted hotels were visited on randomly selected days to collect the
data
. Based on this number, the study population was calculated to be
approximately 2,334 customers who stayed in ‘five-star’ hotels in Jordan.
6. Customers7 were approached randomly, the aim of the research was explained
to them, and then they were invited to participate in the study by completing the
questionnaire. The response rate was 61%, a satisfactory percentage for this kind of
research (Vaus 1999), giving a total number of 556 collected questionnaires.
However, twenty-four cases were noted as outliers8 and have been excluded from the
analysis, as will be explained shortly. The data collection stage took two months to
complete9
4 The Jordanian Hotel Association (2006 statistics). 5 There are 467 classified hotels in Jordan. Thirteen of these are ranked as five-star, seventeen hotels are four-star, and forty-one hotels three-star. The total tourist population was 1,509,320 based on arrival figures. 6 Based a pre-agreement with the hotels, the researcher was allowed to visit the hotels only on specific days. 7 The targeted hotels consisted of five five-star hotels. 8 Outliers are cases with values/scores either above or below the average values/scores by a considerable degree. 9 The data collection started on the 1st of August 2007 and finished by the end of September 2007.
.
As noted above, a total of 556 questionnaires were returned. However, after
eliminating unusable values (n = 6) and respondents who did not provide at least five
answers in the measurement scales (n = 5) as well as deleting outliers (n = 16), a total
of 529 responses were carried forward for further statistical data analyses.
Chapter Seven – Descriptive Analysis of Measurement Scale Items
-177-
Table 7.1: Survey Response Rate
Number Total targeted population 2,335 Total returned questionnaires 556 Unusable observations 6 Missing data 5 Outliers 16 Total used in statistical analysis 529
7.2.1 Questionnaire Fatigue
For the purposes of overcoming questionnaire fatigue, the scale measurement
items were shuffled and mixed randomly; four versions of the questionnaire were
created and distributed in a random fashion. In addition, Chi-square tests were used to
find out if there were different distributions between early respondents and late
respondents10
10 People who completed the questionnaire in August and people who completed it in September.
in terms of their demographic characteristics, and also to distinguish if
there were differences in mean scores between these two groups in terms of the
measurement scale items. These tests were used to determine whether the groups were
statistically different from each other. The results of the Chi-square tests revealed that
there was no difference in distribution between the early respondents and the late
respondents in terms of gender, age group or general behavioural attributes within the
hotel (for example, the type of service used). The results of chi square also showed
that there was no difference between the two groups for the measurement scale items.
As a result, it can be concluded that there was an absence of questionnaire fatigue in
the collected data.
7.2.2 Data Screening
Data screening prior to applying any analytical technique (in this case, SEM)
provided a basic understanding of the relationships between the variables in the study.
Furthermore, the screening ensured that the data underlying the analysis met all of the
requirements for the intended multivariate technique (Hair et al. 2006).
Chapter Seven – Descriptive Analysis of Measurement Scale Items
-178-
Several checks were carried out to screen the data before the analysis. These
included: errors, missing data, normality, Skewness and kurtosis, outliers, reporting
mean (M), standard deviation (SD) and range of scores. These screening methods will
be discussed in detail below.
In addition to the screening methods mentioned above, this thesis
acknowledges non response and its limitation to the research because; non-response
causes bias in surveys by allowing only the views of the people who answered the
survey to be accounted for in the results and this has been defined as “... the bias that
exists when respondents to a survey are different from those who did not respond in
terms of demographic or attitudinal variables” (Sax et al. 2003:411). Some
researchers, for instance, Andrews (2009), have found that non-response bias can
have a negative impact on the research because not all respondents participated in the
study. However, it has been argued that non-response does not always mean a bias, it
depends on how representative of the entire population the non responders are (see for
example, Groves 2006; Cook et al. 2000).
To reduce the non-response bias, this study followed Andrews’ (2009)
recommendations in face-to-face interactions; by taking the points of view of the
respondents into consideration when explaining the purpose of the research to them.
The research team described the questions in the questionnaire and showed how the
results of the study could help service organisations to improve their provided
service(s); hence, making it appear to be mutually beneficial (Andrews 2009). This
has led to a response rate of 61% which is a representative sample according to (Vaus
1999).
7.2.3 Errors
The data screening process followed two steps for error intervention. First,
errors in the data were identified by checking variables for scores that are out of
range. Second, an error treatment process was applied, in which some of the values
were deleted or altered depending on their scores and range; this method was
advocated by Tabachnick and Fidell (2007), who argued that purifying the data prior
to analysis yields clearer results by removing extreme values that might influence the
data negatively.
Chapter Seven – Descriptive Analysis of Measurement Scale Items
-179-
The process of examining errors in the coded data was based on identifying
the values that fell outside the range of possible values for a variable. For instance,
males were giving a value of 1 and females a value of 2; no other values were
possible for this variable, i.e. all answers had to be either 1 or 2. A similar logic was
applied for the other socio-demographic questions and the scale items. All
observations were checked for any values outside the range; this was been done by
conducting frequency analysis for all the variables and examining the outcome in
SPSS (minimum and maximum values, valid and missing values). After this analysis
it can be concluded that the coded data is error-free.
7.2.4 Missing Data
Missing data is a problem that occurs when valid values for one or more
variables are absent (Hair et al. 2006). Missing values did not reach a critical point,
i.e. more than 10% per variable and/or observation11
The first test was the ‘5% Trimmed Mean’ (Tabachnick and Fidell 2007),
which comprises of a comparison between the trimmed mean (that is, a mean which
has 5% trimmed from both of its ends) and the regular mean (a mean which has not
been trimmed) of a variable. This comparison indicates whether there are any
differences between the two means because the trimmed mean should exclude any
extreme values associated with the variable. If both means are close in value, this
. After identifying the nature of
missing values by diagnosing the level of randomness, there were 10 random cases,
which have been considered as ‘Missing at Random’ (MAR). These missing data
were substituted using ‘imputation’ in SPSS to estimate their values.
7.2.5 Normality Tests
Since SEM was used for testing the proposed model in this thesis, any
violation of the assumptions for multivariate normality could invalidate the SEM
statistical results (Byrne 1995). Normality is often used to describe “…a symmetrical,
bell-shaped curve, which has the greatest frequency scored in the middle with smaller
frequencies towards the extremes” (Pallant 2007:57). Based on this definition, the
normality of the data was verified through a series of tests as described below.
11 See Hair et al.’s (2006) argument.
Chapter Seven – Descriptive Analysis of Measurement Scale Items
-180-
indicates that the data form a normal distribution without being biased towards either
end of the scale (Tabachnick and Fidell 2007).
The results of the ‘5% trimmed mean’ test (see Table 7.2), showed that
integrity has an original mean 17.51 and trimmed mean 17.67, benevolence 17.03 and
17.16, competence 16.96 and 17.09, value alignment 16.59 and 16.72, consistency
16.53 and 16.63, communication 16.81 and 16.907, behavioural loyalty 16.92 and
17.03, attitudinal loyalty 17.18 and 17.32, and trustworthiness 16.82 and 16.95. Based
on these results, we can argue that the two mean values (original and trimmed) are not
substantially different for any of the variables. In all cases, the extreme scores in the
sample have less than 1% influence on the original mean, indicating that the data has
a normal distribution without outliers (see Table 7.2).
Table 7.2: 5% Trimmed Mean
Variable Mean % 5% trimmed mean
Integrity 17.51 0.9 17.67
Benevolence 17.03 0.74 17.16
Competence 16.96 0.7 17.09
Value alignment 16.59 0.68 16.72
Consistency 16.53 0.52 16.63
Communication 16.81 0.63 16.907
Behavioural loyalty 16.92 0.8 17.03
Attitudinal loyalty 17.18 0.73 17.32
Trustworthiness 16.82 0.76 16.95
A Kolmogrov-Smirnov test of normality was also conducted. This test
suggests that “a non-significant12
12 The level of significance was .05 (Hair et al. 2007).
result indicates normality” (Pallant 2007:62). In this
case the significant value is 0 for all the scale variables (see Table 7.3), which shows
that there is no violation of this assumption and the data has a normal distribution.
Chapter Seven – Descriptive Analysis of Measurement Scale Items
-181-
Table 7.3: Tests of Normality
Variable
Kolmogorov-Smirnov(a)
Statistic Sig.
Integrity .158 .000
Benevolence .166 .000
Competence .142 .000
Value Alignment .140 .000
Consistency .135 .000
Communication .159 .000
Behavioural loyalty .192 .000
Attitudinal loyalty .144 .000
Trustworthiness .136 .000
After examining several normality tests on the coded data, it can be concluded
that the distribution of scores was reasonably normal, and none of the normality
assumptions have been violated. Therefore, the collected data can be carried forward
and subjected to multivariate data analysis methods.
7.2.6 Outlier Identification
As discussed by several scholars (for example, see Sweet and Martin 2003;
Bryman and Cramer 1999), outliers negatively influence most statistical techniques.
Four tests for identifying outliers in the coded data were utilised:
1. A histogram test: in which the tails of the distribution were checked to identify
potential outliers. Here a total of 31 values from the variables were identified
as outliers, but no action took place until further identification had confirmed
or discarded these values.
2. A boxplot graph: here the values were plotted as circles with numbers attached
(n = 31).
3. A scores check: within this test the scores range was checked to examine
whether any outliers occurred within the range. This is often recommended
with three standards deviations.
4. 5% Trimmed Mean: tested the normality of the data as well as outliers (3.3).
Chapter Seven – Descriptive Analysis of Measurement Scale Items
-182-
After being identified as outliers in the coded sample, 16 observations were
excluded from the analysis. After excluding the outliers, the final observation number
was 529; this forms the body of data for all the statistical analyses to follow (see
Table 7.1).
7.3 Descriptive Findings13
The results from the gender analysis are shown in Table 7.4. 235 (44.4%) of
the customers were males and 294 (55.6%) were females, giving a total of 529
respondents with M = 1.56, SD = .496. there was no statistical differences between
the male and female when applying Chi square test
In the first phase of the descriptive analysis, frequency distributions were
performed on selected sets of socio-demographic data. Several frequency distribution
tests were conducted, for data concerning gender, age, nationality, length of
relationship with the hotel, type of service used, purpose of visit, and length of
relationship with the service provider.
7.3.1 Gender of Respondents
14
Table 7.4: Gender of Respondents
(χ2) .0 >.
Gender Frequency Percent
Male 235 44.4
Female 294 55.6
Total 529 100
7.3.2 Nationality of the Sample
It is evident from Table 7.5 below that the majority of the customers came
from Jordan (42.5%). 14.8% of the customers came from Europe, and 22.7% from the
United States of America (USA). This shows that the majority of the sample is made
up of Jordanian ‘local tourists’ and travellers from USA and Europe. This will allow 13 See appendix eleven for a list of key statistical terminologies and definitions used in this chapter. 14 It was decided, in this thesis, not to apply t-test of the descriptive statistics because t-test is a statistical hypothesis test used when the standard deviation is not really a specific number, or if the data is very small.
Chapter Seven – Descriptive Analysis of Measurement Scale Items
-183-
further investigation of the differences between these groups, since they form the
majority of the sample.
Table 7.5: Nationality of Respondents
Country Frequency Percent Britain 40 7.6 Egypt 22 4.2 France 9 1.7 Germany 13 2.5 India 3 .6 Iraq 14 2.6 Japan 16 3.0 Jordan 225 42.5 Kuwait 4 .8 Lebanon 24 4.5 Spain 16 3.0 Syria 18 3.4 UK (Wales and
Scotland) 5 .9
USA 120 22.7 Total 529 100.0
7.3.3 Age Groups
The age of the participants who were involved in the study ranged from under
20 years old (18% of the sample), 20–25 (31.6%), 26–35 (23.3%), 36–45 (11.7%),
46–55 (10.8%) and 56–65 (4.7%). As can be seen from Table 7.6, the dominant age
groups are those between 20–25 (31.6%) and 26–35 (23.3%) years of age. Therefore,
the customers who participated in this research were mostly young people, who might
perceive trustworthiness differently from the older age groups within the sample. A
Chi-square test was conducted to determine whether any differences existed between
the age groups. The test indicated that there are some significant differences within
the age groups for the sample. These differences are related to the young customers
(20-35), in which they appear to have different needs and expectations compared with
the other age groups in the sample.
Chapter Seven – Descriptive Analysis of Measurement Scale Items
-184-
Table 7.6: Age Groups
Age groups
Frequency
Percent
Cumulative Percent
Under 20 95 18.0 18.0
20–25 167 31.6 49.5
26–35 123 23.3 72.8
36–45 62 11.7 84.5
46–55 57 10.8 95.3
56–65 25 4.7 100.0
Total 529 100.0
7.3.4 Hotel Services
To gain an insight into the attributes and social behaviour of the participants’
additional questions were added to identify the main contact areas and services that
customers use within the hotel, and the associated areas from which trustworthiness
may emerge.
7.3.5 Purpose of the Visit
Customers were asked to state the purpose of their visit in order to understand
the type of stay. This is because, for example, business customers require different
services compared with leisure customers, so that the process of building
trustworthiness may differ between the two groups. As Table 7.7 shows, 51% of the
customers used the hotel for leisure purposes, 32.3% for social events and 16.6% for
business trips.
A χ2 test was conducted to determine whether any difference existed between
customers based on their reason for visiting the hotel. The test indicated that there
were no significant differences in terms of the purpose of visit and, therefore, no
further actions were taken to differentiate between the groups.
Chapter Seven – Descriptive Analysis of Measurement Scale Items
-185-
Table 7.7: Purpose of the Visit
Purpose Frequency Percent
Cumulative
Percent
Business 88 16.6 16.6
Leisure 270 51.0 67.7
Social event 171 32.3 100.0
Total 529 100.0
7.3.6 Type of Used Service
Building on the previous section (7.3.5), a further breakdown of the areas that
customers used in the hotel was carried out to facilitate an understanding of the main
points of customer interaction with the hotel, thereby investigating whether the
formation of trustworthiness occurs within a particular area(s).
From Table 7.8 below, it can be seen that the highest percentage of visiting
customers (34%) visit the hotel mainly for its bar/coffee shop, which is not entirely
surprising since most participants came from Jordan and they are normally do not
require accommodation. This is also can be seen when the purpose of the visit is
compared with the service used, as shown in Table 7.9.
Table 7.8: Type of Used Service
Services Used Frequency Percent Cumulative Percent
Accommodation 94 17.8 17.8
Restaurant 58 11.0 28.7
Conference Room 74 14.0 42.7
Bar/Coffee shop 180 34.0 76.7
Health Club 98 18.5 95.3
Banqueting 25 4.7 100.0
Total 529 100.0
Chapter Seven – Descriptive Analysis of Measurement Scale Items
-186-
Table 7.9: Service Used vs. Purpose of Visit
Type of Service
Purpose of Visit
Business Leisure Social event
Accommodation 15 35 44
Restaurant 10 30 18
Conference Room 22 21 31
Bar/Coffee shop 5 134 41
Health Club 24 39 35
Banqueting 12 11 2
Total 88 270 171
7.3.7 Longevity of Customer Relationships
The length and frequency of the visits point toward the strength of the already
established relationship between the hotel and its customers (see Tables 7.10 and
7.11). 27% of the participants have less than one year’s relationship with the hotel,
15.1% have had a relationship lasting 1–2 years, 9.5% lasting 3–5 years, 24.2%
lasting 6–10 years, 20% lasting 11–20 years, and 4.2% have had a relationship lasting
over 20 years. This shows that the majority of customers 44.2% have a relatively long
relationship with their hotel.
A Chi-square test was carried out to determine whether any differences existed
between customers in terms of their relationship with the hotel. The test showed that
there are some significant differences between the groups particularly when
comparing short relationships (less than one year) with long relationships (over 20
years). However, it is considered in this thesis, that these significant differences may
prove useful when investigating each relationship individually because each group is
perceived to have unique characteristics that are different than the other groups;
hence, it allows greater understanding of each group and how the relationship can be
enhanced and managed. This is, however, beyond the scope of this thesis.
Chapter Seven – Descriptive Analysis of Measurement Scale Items
-187-
Table 7.10: Length of the Relationship between Customers and the Hotel Frequency Percent Cumulative Percent
Less than 1 year 143 27.0 27.0
1–2 years 80 15.1 42.2
3–5 years 50 9.5 51.6
6–10 years 128 24.2 75.8
11–20 years 106 20.0 95.8
Over 20 years 22 4.2 100.0
Total 529 100.0
Table 7.11: Frequency of Service Usage Frequency Percent Cumulative Percent
Once a year 92 17.4 17.4
1–2 times a year 134 25.3 42.7
3–5 times/year 168 31.8 74.5
6/10 + times/year 135 25.5 100.0
Total 529 100.0
7.4 Means and Standard Deviation Analysis
The results of descriptive statistics analyses for the scale items that measured
trustworthiness are presented in Table 7.12. The measurement scale consists of 36
items, aiming to measure the determinants of trustworthiness (competence,
consistency, integrity, benevolence, communication and value alignment),
trustworthiness (the model’s focal point), and attitudinal and behavioural loyalty
(model outcome). This illustrates the fact that each variable is measured by four
items, which is acceptable for SEM (Hair et al. 2006; Byrne 1995). Respondents were
asked to provide answers on each item measured by a five-point Likert scale, ranging
Chapter Seven – Descriptive Analysis of Measurement Scale Items
-188-
from 1 (‘Strongly Disagree’) to 5 (‘Strongly Agree’), with 3 as a ‘Neutral’ point in the
scale.
The mean of a data set is the arithmetic average of the values in the set,
obtained by summing the values and dividing by the number of values. Therefore, the
mean is a measure of the centre of the distribution and is a weighted average of the
item scores, with the relative frequencies as the weight factors.
The standard deviation measures of the spread of the distribution about the
mean. Standard deviation measures spread in the same physical unit as the original
data, by calculating the square root of the variance.
Based on the mean score of each item15
Model constructs
, respondents, for example, tended to
strongly agree that the hotel always keeps its word (M = 4.42, SD = .67) and
communicates clearly (M = 4.22, SD = .79). Moreover, they also agreed that they will
say positive things to other people about the services provided at the hotel (M = 4.23,
SD = .78), and they will continue visiting the hotel even if the service charges are
increased moderately (M = 4.34, SD = .70).
Furthermore, respondents agreed that the hotel is always honest with them (M
= 4.17, SD = .81), it responds caringly when they share their problems (M = 4.2, SD =
.77), and overall respondents agreed that the hotel is trustworthy (M = 4.31, SD =
.75). A more detailed discussion will follow below.
7.4.1 Proposed Model Constructs
Table 7.12: Mean and Standard Deviation for the Model Constructs
Construct Mean Std. Deviation
Low level determinants* 33.50 3.35 High level determinants** 67.94 4.72 Integrity 17.51 2.02 Benevolence 17.03 2.32 Value alignment 16.58 2.54
15 The results of standard deviation and score mean for the items are reported later on in this chapter.
Chapter Seven – Descriptive Analysis of Measurement Scale Items
-189-
Consistency 16.53 2.51 Competence 16.97 2.33 Communication 16.81 2.38 Behavioural loyalty 16.92 2.46 Attitudinal loyalty 17.18 2.17 Trustworthiness 16.82 2.37
*combined values of: consistency and competence **combined values of: integrity, benevolence, value alignment, communication
As can be seen from Table 7.12, the high level determinants of trustworthiness
scored significantly higher than low level determinants. This is because two particular
constructs, integrity and benevolence, appear to have higher scores compared with the
rest of the constructs. There is also strong evidence for this phenomenon in the
literature as a characteristic of trust (Larzelere and Huston 1980; Butler and Cantrell
1984; Rempel 1985). However, integrity is viewed here as a determinant of
trustworthiness this is in line with the work of Sheppard and Sherman (1998) and
Mayer et al. (1995).
It should also be noted that attitudinal loyalty scored higher than behavioural
loyalty, which is an indication of high level involvement with the service provider
(Rundle-Thiele and Bennett 2001).
7.4.2 Integrity
Table 7.13 shows how respondents evaluated the four scale items related to
the construct of integrity.
Table 7.13: Mean and Standard Deviation for the Integrity Construct
Integrity
Item Mean Std. Deviation
My … show fairness in transactions 4.32 .70
My … always achieves consistency in
service delivery
4.33 .67
Chapter Seven – Descriptive Analysis of Measurement Scale Items
-190-
My … always keeps its word 4.42 .67
The …. employees treat me with respect 4.42 .67
The analysis in Table 7.13 shows that sample members rated the item ‘The …
employees treat me with respect’ as the most significant item for integrity, followed
by ‘My … always keeps its word’. These items are related to higher level
determinants of trustworthiness, because they indicate a selfless characteristic in the
service provider.
7.4.3 Benevolence
Table 7.14 shows how respondents evaluated the four scale items related to
the construct of benevolence.
Table 7.14: Mean and Standard Deviation for the Benevolence Construct
Benevolence
Item Mean Std. Deviation
My … is open to my needs 4.22 .73
My … acts in a caring manner 4.32 .75
My … keeps its customers’ best interest
in mind during most transactions
4.20 .78
My … is receptive to my needs 4.27 .75
The analysis in Table 7.14 indicates that the sample members rated the items
‘My … acts in a caring manner’ and ‘My … is receptive to my needs’ as the most
positive items for benevolence. This is a similar result to that for the integrity items
listed above, since these items also represent altruistic behaviour from the service
provider.
7.4.4 Competence
Table 7.15 demonstrates how respondents evaluated the four scale items
related to the construct of competence.
Chapter Seven – Descriptive Analysis of Measurement Scale Items
-191-
Table 7.15: Mean and Standard Deviation for the Competence Construct
Competence
Item Mean Std. Deviation
My … has adequate skills to deliver the
right service
4.22 .73
My … is efficient 4.28 .75
The … employees can competently
handle most of my requests
4.19 .77
The … employees can be relied upon to
know what they are doing
4.27 .75
It is evident from Table 7.15 that the sample members graded the item ‘My …
is efficient’ most highly within the competence construct, followed by ‘the …
employees can be relied upon to know what they are doing’. This suggests that,
despite that fact that customers are concerned with the high level determinants of
trustworthiness, they still require the service provider to be efficient.
7.4.5 Value Alignment
Table 7.16 illustrates how respondents evaluated the four scale items related to
the construct of value alignment.
Table 7.16: Mean and Standard Deviation for the Value Alignment Construct
Value alignment
Item Mean Std. Deviation
My … shares the same values as me 4.09 .76
My … is ethical while dealing with me 4.25 .82
My … is interested in more than just
making profit out of customers
4.07 .82
The employees act as if they value the
customer
4.16 .80
The analysis in Table 7.16 indicates that respondents rated the item ‘My … is
ethical while dealing with me’ significantly higher than the rest of the items related to
the construct of value alignment. This can be related to the growing interest from both
academic and professional researchers regarding ethical considerations in business
Chapter Seven – Descriptive Analysis of Measurement Scale Items
-192-
transactions. However, this topic is beyond the scope of this research; it will be
addressed in the future research section in the concluding chapter of this thesis.
7.4.6 Consistency
Table 7.17 demonstrates how respondents evaluated the four scale items
related to the construct of consistency.
Table 7.17: Mean and Standard Deviation for the Consistency Construct
Consistency
Item Mean Std. Deviation
My … has stability in its operations 4.09 .83
My … is always reliable 4.14 .79
My … always matches customers’
expectations
4.10 .83
My … always satisfies customers’ needs 4.19 .81
It is evident from the results in Table 7.17 that the sample members graded the
item ‘My … always satisfies customers’ needs’ most positively in relation to
consistency, followed by ‘My … is always reliable’. This shows that satisfaction is
still important to customers as a low level determinant of trustworthiness, as well as
the higher level determinants mentioned above.
7.4.7 Communication
Table 7.18 indicates how respondents evaluated the four scale items related to
the construct of communication.
Table 7.18: Mean and Standard Deviation for the Communication Construct
Communication
Item Mean Std. Deviation
My … communicates clearly 4.22 .79
The … respond immediately when
contacted
4.28 .70
My … inform me immediately of any
problems
4.12 .77
Chapter Seven – Descriptive Analysis of Measurement Scale Items
-193-
The … always communicates with me 4.17 .77
The analysis in Table 7.18 shows that respondents rated the items ‘the …
respond immediately when contacted’ and ‘My … communicates clearly’
significantly higher than the rest of the items related to the construct of
communication. This suggests that communication is an important issue, with
customers expecting an immediate response from the hotel when they are contacted
regarding an extra service or a complaint. However, customers are not only concerned
with an immediate response; they also require clarity of communication, and an
assurance that their problem(s) have been fully addressed.
7.4.8 Behavioural Loyalty
Table 7.19 shows how respondents evaluated the four scale items related to
the construct of behavioural loyalty.
Table 7.19: Mean and Standard Deviation for the Behavioural Loyalty Construct
Behavioural loyalty
Item Mean Std. Deviation
I will transact with this … again for
future visits
4.22 .75
I will try new services that are provided
by this …
4.30 .72
I will recommend other people to visit to
this …
4.16 .77
I will say positive things to other people
about the services provided at this …
4.23 .78
It is clear from Table 7.19 that the sample respondents rated the item ‘I will try
new services that are provided by this …’ the most highly with regard to consistency,
followed by ‘I will say positive things to other people about the services provided at
Chapter Seven – Descriptive Analysis of Measurement Scale Items
-194-
this …’. This is an indication that once behavioural loyalty is established, customers
show willingness to recommend the service provider by word of mouth, and at the
same time they show confidence in trying new services offered by the service
provider because at this stage they perceive the service provider to be trustworthy.
7.4.9 Attitudinal Loyalty
Table 7.20 shows how respondents evaluated the four scale items related to
the construct of attitudinal loyalty.
Table 7.20: Mean and Standard Deviation for the Attitudinal Loyalty Construct
Attitudinal loyalty
Item Mean Std. Deviation
I will continue to visit this … even if the
service charges are increased moderately
4.34 .70
I have strong loyalty to this … 4.33 .66
I will keep visiting this … regardless of
everything being changed somewhat
4.25 .74
I am likely to pay a little bit more for
using the services of this …
4.25 .73
The analysis in Table 7.20 shows that the sample respondents rated the items
‘I will continue to visit this … even if the service charges are increased moderately’
and ‘I have strong loyalty to this …’ significantly higher than the rest of the items
related to the construct of attitudinal loyalty. This represents the attitudinal
attachments between customers and the service provider, by which they have
established strong ties. These strong relationships are resilient in the face of many
influential factors – such as price, clearly demonstrated by the positive response to the
first item.
Chapter Seven – Descriptive Analysis of Measurement Scale Items
-195-
7.4.10 Trustworthiness
Table 7.21 illustrates how respondents evaluated the four scale items related to
the construct of trustworthiness.
Table 7.21: Mean and Standard Deviation for the Trustworthiness Construct
Trustworthiness
Item Mean Std. Deviation
The … employees are professional and
dedicated to customers
4.13 .77
The … respond caringly when I share
my problems
4.20 .77
My … is always honest with me 4.17 .81
Overall I feel I can trust my … 4.31 .75
It is evident from the results in Table 7.21 that the sample respondents rated
the items ‘The… respond caringly when I share my problems’ and ‘Overall I feel I
can trust my …’ as the two most positive items related to the trustworthiness
construct. These results signify that the service provider has to act in a trustworthy
manner in order to gain customers’ trust; this is achieved by applying all the previous
antecedents at both the construct and item levels.
The above analysis only constitutes a general overview of the individual scale
items, and provides a simple descriptive analysis to establish an initial understanding
of these items and their relative importance within their respective constructs. The
scale items will be further analysed and tested in the discussion of path analysis in the
next chapter.
Chapter Seven – Descriptive Analysis of Measurement Scale Items
-196-
7.5 Additional Influential Factors
From Table 7.22 below we can conclude that the service provider’s brand
name is an important factor for both male and female participants (responses indicated
that 208 strongly agreed and 226 agreed with the statement). Building a strong brand
for the hotel improves the chances of building trust between service provider and
customer. In this context the role of the international hotel chain is crucial, because
the majority of five- and four-star hotels in Jordan are part of international chains. It is
also the case that applying high security standards increases the chance of customers
trusting the service provider (Table 7.23).
Table 7.22: Brand Name vs. Gender
The hotel brand name is very important
Gender Total
Male Female
Strongly Disagree 6 0 6
2 21 7 28
Neutral 22 39 61
4 105 121 226
Strongly Agree 81 127 208
Total 235 294 529
Table 7.23: Security of the Hotel vs. Gender
I trust the hotel more if the hotel
applies higher security standards Gender Total
Male Female
Strongly Disagree 0 4 4
2 25 6 31
Neutral 32 50 82
4 96 78 174
Strongly Agree 82 156 238
Total 235 294 529
Chapter Seven – Descriptive Analysis of Measurement Scale Items
-197-
From the results from Table 7.22 and 7.23 we can argue that there are other
factors contributing to the creation trust in a hotel context. These include higher
security, atmosphere and brand name. However, none of these factors have been
investigated in the existing literature surrounding trustworthiness. Therefore, further
investigation will be suggested in the recommendations chapter, in order to explore
the extent of the influence of these factors on trustworthiness within the hotel sector.
7.6 Reliability and Validity of Measurement Scales
7.6.1 Reliability of Measurement Scales
Scale reliability is an essential factor in any latent measurement scale
(Netemeyer et al. 2003). Reliability “is concerned with that portion of measurement
that is due to permanent effects that persist from sample to sample” (Netemeyer et al.
2003:10). Reliability is often measured by an internal consistency test; this assessment
of reliability shows the synchronisation of items comprising a measurement scale. The
meaning of internal consistency is the extent to which its items are inter-correlated.
High inter-item correlations indicate that the items in a scale have a strong
relationship to the latent construct, and are probably measuring the same thing
(Tabachnick and Fidell 2007).
The internal consistency of a measurement scale is assessed by Cronbach’s
coefficient alpha (α). Calculating Cronbach’s alpha, along with the item-to-total
correlation for each item, forms an examination of the overall reliability of the
measurement scale (Tabachnick and Fidell 2007). It is accepted that if a measurement
scale has a Cronbach’s α coefficient above .70; it is thought to be an internally
consistent scale so that further analysis is possible. Conversely, if the scale has an
alpha coefficient lower than .70 the scale should be investigated for any sources of
measurement errors, such as insufficient sampling of items, administration errors,
situational factors, sample characteristics, number of items, and conceptual errors in
formulating the measurement scale (Netemeyer et al. 2003).
As an initial examination of the reliability of the measurement scales for the
nine constructs of the proposed model, Cronbach’s alpha coefficients were computed
in SPSS 16.0 and are presented in Table 7.24. All of the measurement scales for the
Chapter Seven – Descriptive Analysis of Measurement Scale Items
-198-
nine constructs achieved an acceptable level of Cronbach’s alpha (above .70),
indicating that the scales are reliable and suitable for further data analysis (Hair et al.
1998).
Table 7.24: Reliability Analysis16
Variable
Number of
Items
Cronbach’s
Alpha (α)
Integrity 4 .73
Benevolence 4 .76
Competence 4 .77
Value alignment 4 .80
Consistency 4 .76
Communication 4 .78
Behavioural loyalty 4 .82
Attitudinal loyalty 4 .76
Trustworthiness 4 .75
Since the integrity construct displays the lowest value for alpha, there was an
attempt to increase its reliability. However, it was discovered that alpha decreased
from .73 to .5, .61 and .45 on deletion of the first, second and third items. Therefore, it
was decided to maintain all the items with no alteration, accepting that a Cronbach’s
alpha of .73 is adequate to form a reliable scale.
The reliability and variance extracted for all the constructs will be reported in
the following chapter. Composite reliability refers to a measure of the internal
consistency of indicators for the construct, illustrating the level to which they support
the corresponding latent construct (Hair et al. 1998). A commonly used value for the
acceptance level for composite reliability is .70. If the calculated reliability is higher
than .70, the indicators for the latent construct are reliable and are measuring the same
construct. As a complementary measure of composite reliability, the variance
extracted can be estimated to explain the total amount of variance in the indicators
that is accounted for by the corresponding latent construct. A commonly used 16 Appendix thirteen provides the full reliability analysis, including; Cronbach's alpha if item deleted, corrected item-total correlation and scale variance if item deleted.
Chapter Seven – Descriptive Analysis of Measurement Scale Items
-199-
threshold of acceptability is .50. If the variance extracted values are high, the
indicators are accepted as being representative of the latent construct.
7.6.2 Validity of Measurement Scales
As it has been discussed in Chapter Five – Section 5.9; reliability of the
measurement scale is related to how dependable a set of items are, validity is
associated with whether a particular construct is the primary cause of item co-
variation (Netemeyer et al. 2003; Tabachnick and Fidell 2007). Validity often
concerns the extent to which the measurement items or indicators measure what they
are intended to measure (Hair et al. 1998). In particular, construct validity deals with
the capability of a scale as a measure of a specific variable.
Content validity provides evidence of judgmental validity. To obtain empirical
evidence, after the measurement scale is distributed the relationships between the
items within the measurement scale are examined, as well as relationships between
the items and constructs and between the constructs. Empirical evidence for validity
can be obtained by calculating criterion-related validity and construct validity. In
order to verify content validity, the measurement scales for the constructs were
examined via a three step process as described in the pilot chapter. Through this
procedure, the content validity of the measurement scales was ascertained by rooting
the model’s constructs with the related literature. Moreover, face validity was also
achieved by examining the research instrument from the perspectives of the three
judges17
For criterion-related validity, as discussed earlier in this thesis
. Consequently translation validity was accomplished in this thesis.
18
17 See Chapter Six - Section 6.3. 18 See Chapter Five - Section 5.9.1.
, concurrent
validity analysis was performed to test whether there was a correlation between the
criterion variables and the measurement scales. The results of the linear regression
analysis will be explored in detail in relation to the results of the SEM procedure.
Nomological and convergent validity analysis will be reported in the next chapter,
along with the results of CFA, in view of the fact that CFA can provide empirical
evidence of construct validity.
Chapter Seven – Descriptive Analysis of Measurement Scale Items
-200-
Discriminant validity refers to a measure of the indicators of different
constructs that theoretically and empirically should not be associated with each other
(Hair et al. 1998). Accordingly, the indicators that measure one construct should not
be correlated with the indicators that measure another construct if the constructs have
discriminant validity. This can be assessed by examining χ2 for every possible pair of
estimated constructs.
Summary
This chapter examined the characteristics of the research sample. The socio-
demographic data of respondents illustrated that the majority of respondents fall
within two main age groups (20–25 years (31.6%) and 26–35 years (23.3%)).
Respondents have had a relatively long relationship with the hotel (an average of 3–6
years), and come from three main destinations (Jordan – 42.5%, Europe – 14.8% and
USA – 22.7%). Following the socio-demographic analysis, the statistical analysis of
the questionnaire scale items will be presented. Data was analysed using SPSS. In the
next chapter, the SEM procedure will be reported to analyse the overall fit and
viability of the proposed model.
Chapter Eight – Multivariate Data Analysis
-201-
CHAPTER EIGHT: MULTIVARIATE DATA
ANALYSIS AND MODEL VALIDATION
8.1 Introduction
The Structural Equation Modelling (SEM) procedure encompasses many
statistical techniques, including regression, factor analysis and path analysis (Hair et
al. 2006). According to Schumacker and Lomax (2004), the original concept for this
type of modelling was developed in the early 1930s by Wright in 1934. He derived a
mathematical way of describing correlations and theories regarding causal
relationships. These relationships were illustrated as path diagrams, leading to the
term path analysis (Schumacker and Lomax 2004). The purpose of this chapter is to
provide a full discussion on the results of SEM, its validation, reliability and how it
impacts the proposed model of this study.
Jöreskog (1973) noted the reintroduction of SEM in the early 1970s, as a
combination of path analysis and factor analysis that enabled the examination of
causal propositions by comparing the covariance of an original data set with that of a
hypothesised model (Iriondo et al. 2003). The essential null hypothesis1
In general terms, SEM allows for the examination of hypotheses concerning
relationships between observed and latent variables. It is in the use of latent variables
that the strength of this statistical technique is demonstrated. Latent variables are
theoretical concepts that connect phenomena under a particular term, and which can
is that the
covariance of an initial data set is equal to the covariance of the model, and in contrast
to standard statistical techniques, the objective is to accept the null hypothesis (Hoyle
1995). In other words, if the theoretical framework provides a good fit for the data,
and the parameters within the model have been estimated with no significant error, the
model covariance will be equal to that of the original data, and the model should not
be discarded (Hoyle 1995), this is discussed in section 8.2.
1 Appendix eleven provides definitions of key statistical terms used in this chapter.
Chapter Eight – Multivariate Data Analysis
-202-
be expressed in terms of directly measured variables (Hoyle 1995). Using this
technique it is possible to construct theoretical concepts and test their reliability, as
well as hypothesis and test theories about their relationships (Byrne 2001).
In order to proceed with a SEM analysis, this chapter builds on the theory
established in the previous chapters, arguing that every variable within the current
model is conceptualised as a latent variable (because they cannot be observed
directly), measured by four items (or indicators). The SEM model in this study
contains two related models and proceeds through two steps; measurement model
validation, and structural model fitting. The items are distinguished by their factor
loading onto latent variables, and which of these latent constructs predict others was
examined using AMOS 7.02
This chapter interprets and discusses the results of the SEM process for the
hypothesised model testing (see Figure 8.1). It examines the model from two
perspectives; the measurement model, and the structural model. The measurement
model estimates the latent variables and associates them with their respective
measured variables. Three points will be examined closely: 1) the measured variables,
and to what extent they reflect the latent variables; 2) the relations between the
observed variables and 3) the reliability of each measured variable. Following this
stage, the structural model will be tested to distinguish how well it fits the data. This
, a software package specifically designed to perform
these statistical techniques.
The measurement model defines the constructs (latent variables) that the
model will exploit, and assigns observed variables to each one. Furthermore, the
measurement model employs Confirmatory Factor Analysis (CFA) to measure the
degree to which the observed variables load on their latent constructs (Bollen 1989).
The structural model then defines the causal relationship between these latent
variables (see Figure 8.2 below; the arrows between the latent variables represent
these structural connections), in order to carry out a path analysis using the latent
variables (Bollen 1989).
2 LISREL, AMOS, and EQS are three popular statistical packages for conducting SEM. The first two are distributed by SPSS. The reason why AMOS has been utilised in this thesis is that fact that AMOS (Analysis of Moment Structures) has a more user-friendly graphical interface than LISREL or EQS (Kline 1998), and it’s the standard package at Coventry University.
Chapter Eight – Multivariate Data Analysis
-203-
will involve an analysis of the paths among the latent variables (the path coefficients).
It is hopes that the outcome of the SEM analysis will be robust enough to allow
general conclusion to be drawn.
Chapter Eight – Multivariate Data Analysis
-204-
Defining the Individual Constructs
What items are to be used as measured variables?
Develop and Specify the Measurement Model Construct measured variables using constructs
Draw path diagram for measurement model
Designing Study to produce Empirical Results Assess the adequacy of the sample size
Select the estimation method and missing data approach
Assessing Measurement Model Validity Assess line of goodness of fit and construct validity
of measurement model
Measurement Model Valid?
Proceed to test structural model
with stages 5 and 6
Refine measures and design a new study
Specify Structural Model Convert measurement model to structural model
Assess Structural Model Validity Assess the goodness of fit and significance,
direction and size of structural parameter estimates
Structural Model Valid?
Draw substantive conclusions and
recommendations
Refine model and test new data
Yes
Yes No
No
Stage 1
Stage 4
Stage 3
Stage 2
Stage 5
Stage 6
(Source: Adapted from Hair et al. 2006:759)
Figure 8.1: The Process of SEM Analysis in This Chapter
Chapter Eight – Multivariate Data Analysis
-205-
8.1.1 Analysis Strategy
This section details the data analyses strategies adapted to test the causal
relationships in the model (the hypotheses), developed in Chapter Four section 4.6
and summarised in Table 8.1. A detailed discussion and examination of the results
will follow in each respective section of this chapter.
Table 8.1: Overview of the Analyses Strategy
Step
New
measure
development
Exploring
underlying
relationships
Existing
measures
validation
Testing of the
conceptual model
Hypotheses
reporting
Main aims
Measure
development,
purification
and
validation
Explore the
data set to
identify
causal
relationships
Measure
validation
Assessment of
theoretically-derived
model
Assessment of the
strength of
relationships between
individual constructs
To draw
conclusions
based on the
various
performed tests
Statistical
methods
employed
Internal
Consistency,
CFA
EFA
CFA
SEM
Sections in this chapter
reporting the analysis strategy
Section 8.2
Section 8.3.1
Section8.3.2
Section 8.3.5
Section 8.6
Chapter Eight – Multivariate Data Analysis
-206-
8.2 The Measurement Model
As noted earlier, each construct in the measurement model can be examined
through a process of CFA. CFA is employed to test the measurement model, defining
the stated relationships between the observed variables and the underlying constructs
(Hair et al. 2006). This approach examines whether the collected data are consistent
with a very constrained hypothesised model, and it is applied before the specified
model stage (Byrne 2001). Therefore, CFA allows the identification and clustering of
the observed variables in a pre-specified theory-determined hypothesised model in
order to evaluate to what extent a particular collected data set confirms what is
theoretically believed to be its underlying constructs (Hair et al. 2006; Bollen 1989).
In this study, the measurement model underlying the hypothesised constructs
was proposed and tested. As the detailed theoretical and empirical aspects of the
construct and the observed indicators were discussed in Chapters Two, Three and
Four, all of the measurement models’ observed and unobserved variables were
developed on the basis of conceptual, theoretical, and empirical research (see Figure
8.2).
During the process of CFA, the measurement model was confirmed in terms
of the measurement of the underlying constructs. Since CFA is performed on the
principle that the observed variables are perfect indicators for the underlying
constructs, all the constructs in the measurement model were tested simultaneously.
As discussed in Chapter Four section 4.6, the model estimation process for the
measurement model will be provided along with the statistical results. A number of
indicators were used to estimate the goodness of fit of the measurement model, these
were; goodness-of-fit index (GFI); adjusted goodness-of-fit index (AGFI);
comparative fit index (CFI); relative fit index (RFI); Bentler Bonett index or normed
fit index (NFI); parsimony normed fit index (PNFI); root mean square residual
(RMR); parsimony goodness-of-fit index (PGFI); incremental fit index (IFI); and root
mean square error of approximation (RMSEA).
Chapter Eight – Multivariate Data Analysis
-207-
In evaluating the CFA, the covariance matrices based on Product Moment
Correlation and Standard Deviation were utilised as input data matrices to examine
the data, with the proviso that the analysis of correlation matrix data may generate
problematic results (Byrne 2001). Additionally, Maximum Likelihood (ML) was used
as a method of parameter estimation because:
1) the collected usable sample was large (n=529);
2) the distribution of the observed variables is multivariate normal;
3) the hypothesised model is assumed to be valid; and,
4) the scales of the observed indicators were continuous (Bollen 1989; Byrne
2001).
The assumption of normal distribution in the observed variables was met
according to an analysis of skewness and kurtosis, and the variables in the
hypothesised model were believed to be valid (see Chapter Seven, section 7.2.5).
Benevolence
Competence
Integrity
Consistency
Value alignment
Communication
Behavioural loyalty
Attitudinal loyalty
Hotel’s Trustworthiness
Low level
High level
+
+
+
+ +
+
+
+ +
Figure 8.2: The Hypothesised Model and the Relationships between Constructs
Chapter Eight – Multivariate Data Analysis
-208-
8.2.1 Testing the Measurement Model
This study began with the development of a theoretical model with clear
relationships between the latent constructs and their associated (observable and
therefore measurable) items. Supporting theories and discussion of the measurement
items and their association with their respective constructs were provided earlier in
this thesis, in the literature review and the chapters on methodology and
conceptualisation. Following this model development stage, the structural model
detailing how the constructs are interrelated with each other was defined by the
proposed hypotheses. Subsequently, it was confirmed that SEM could be used to test
the proposed hypotheses.
In SEM, the development of the hypothetical model describing the linkages
between the latent constructs and their empirical observed indicators is seen as the
measurement model, while the theoretical relationships between the constructs is
referred to as the structural model (Bollen 1989; Byrne 1998). The measurement
model can specify the patterns of how the observed indicators load on the constructs,
and it also provides the measurement attributes of reliability and validity for the
observed indicators.
During the SEM process, once the necessary information has been gathered
and the requirements of the full structural model are met, the exogenous
(independent) and endogenous (dependent) constructs can be defined. Changes in the
values of the exogenous constructs are not explained by the model, but changes in the
values of the endogenous constructs are influenced by the exogenous constructs. All
the constructs fall into one of these two categories.
Based on the above discussion, nine theoretical constructs were discussed in
terms not only of their posited relationships with the observed indicators, but also
with regard to the structural relationships between the constructs (see Figure 8.3 and
Figure 8.4). as shown in Table 8.2, the constructs are trustworthiness (TW),
consistency (CO), competence (CP), integrity (IN), benevolence (BN), value
alignment (VA), communications (CM), behavioural loyalty (BL) and attitudinal
loyalty (AL).
Chapter Eight – Multivariate Data Analysis
-209-
Table 8.2: The Model Indicators (items) in Relation to the Model Constructs
Constructs and Items3 Integrity (IN) Communication (CM) IN1: Show fairness in transactions CM1: Communicates clearly IN2: Always achieves consistency in service delivery
CM2: The ….. respond immediately when contacted
IN3: Always keeps its word CM3: Inform me immediately of any problems
IN4: The ….. employees treat me with respect
CM4: The ….. always communicates with me
Benevolence (BN) Behavioural Loyalty (BL) BN1: Is receptive to my needs BL1: I will transact with this ….. again for
future visits BN2: Acts in a caring manner BL2: I will try new services that are provided
by this ….. BN3: Keeps its customers’ best interest in mind during most transactions
BL3: I will recommend other people to visit to this …..
BN4: Is receptive to my needs BL4: I will say positive things to other people about the services provided at this …..
Competence (CP) Attitudinal Loyalty (AL) CP1: Has adequate skills to deliver the right service
AL1: I will continue to visit this ….. even if the service charges are increased moderately
CP2: Is efficient AL2: I have strong loyalty to this ….. CP3: The ….. employees can competently handle most of my requests
AL3: I will keep visiting this ….. regardless of everything being changed somewhat
CP4: The ….. employees can be relied upon to know what they are doing
AL4: I am likely to pay a little bit more for using the services of this …..
Value Alignment (VA)
Trustworthiness (TW)
VA1: Shares the same values as me TW1: The ….. employees are professional and dedicated to customers
VA2: Is ethical while dealing with me TW2: The ….. respond caringly when I share my problems
VA3: Is interested in more than just making profit out of customers
TW3: My ….. is always honest with me
VA4: The ….. employees act as if they value the customer
TW4: Overall I feel I can trust my …..
Consistency (CO) CO1: Has stability in its operations CO2: Is always reliable CO3: Always matches customers’ expectations
CO4: Always satisfies customers’ needs 3 As noted previously (Chapter Six, Section 6.4.6.2), a five-point Likert scale was adopted to measure the items, in which: 1 = Strongly disagree, 3 = Neutral and 5 = Strongly agree.
Chapter Eight – Multivariate Data Analysis
-210-
Figure 8.34
illustrates the first part of the proposed model (determinants of
trustworthiness). It shows the relationships between the model’s constructs, and how
they are correlated with each other and it demonstrates the associated items with each
construct to provide greater clarity in terms of how these items are linked to the
constructs.
4 ‘E’ refers to the standard error for each item
Competence
Consistency
Benev.
Integrity
Value alignment
Comm.
CO3
CO4
CP1
CP2
CP3
IN4
VA1
VA2
VA3
VA4
CM1
CM2
CM4
BN1
CO2
CO1
IN3
IN2
IN1
CP4
CM3
BN2
BN3
BN4
E3
E4
E5
E6
E7
E12
E13
E14
E2
E1
E11
E10
E9
E8
E15
E16
E17
E18
E20
E21
E19
E22
E23
E24
Figure 8.3: Part One of the Hypothesised Measurement Model
Chapter Eight – Multivariate Data Analysis
-211-
The second part of the model (the model outcome) is illustrated in figure 8.4
where the related items are linked to their constructs. Both parts of the model (Figure
8.3 and 8.4) will be explored in details in the following section.
8.2.2 Overall Measurement Model
The measurement model for the constructs and the observed indicators was
determined on the basis of the statistical and theoretical dependability of the
constructs. Therefore, each final model represents the initial model without alteration
or adjustment with regard to their parsimony and essential meaningfulness.
In this context, thirty-six observed indicators, associated with nine constructs,
were determined using CFA, as shown in Figures 8.3 and 8.4 and Table 8.2. The
overall measurement model consisted of nine constructs (see Figures 8.3 and 8.4),
with four items loading onto each construct.
This study followed a ‘strictly confirmatory approach’ as specified by Hoyle
(1995), by which a model is tested using SEM goodness-of-fit tests to verify whether
the pattern of variances and covariances in the data is consistent with a structural
model. A cross validation method was used to check the validity of the sample;
despite the fact that this method is most commonly used in the ‘model development
approach’, it has been used here for the purposes of clarification.
Behavioural loyalty
Attitudinal loyalty
Trustworthiness
AL4
AL3
AL2
AL1
BL1
BL2 TW4
TW3
TW2
BL3
BL4
TW1
E28
E27
E26
E25
E29
E30
E31
E32
E33
E34
E36
E35
Figure 8.4: Part Two of the Hypothesised Measurement Model
Chapter Eight – Multivariate Data Analysis
-212-
As mentioned in Chapter Five - Section 5.10.2.1, the literature surrounding
SEM debates the use of one-step modelling (Hayduk 1987; Fornell and Yi 1992) as
opposed to two-step modelling (see for example, Anderson and Gerbing 1988)5
Consequently, this study tested the replication of the overall measurement
model, with nine constructs and thirty-six observed indicators, across the first sample
(n = 265) and the second sample (n = 264). The first sample will be referred to as the
. In
the two-step modelling approach, researchers initially test the pure measurement
model underlying a full structural equation model, and if the fit of the measurement
model is acceptable, they then proceed to the second step of testing the structural
model by comparing its fit with that of different structural models (for example, the
null model). Others have argued that SEM researchers should apply a four-step
modelling approach to allow a great control over the model fit (see for instance,
Mulaik and Millsap 2000). This study follows a two-step approach, where the
measurement model will be examined and after attaining acceptable results the
structural model will be accepted. This method is used by the leading marketing
academics and it permits the model to be altered (if needed) in order to have a better
fit (see for example, Garbarino and Johnson 1991; Morgan and Hunt 1994)
In applying SEM, numerous approaches for cross validation have been
introduced, depending upon the objective of the study (Anderson and Gerbing 1988).
The split sample approach is commonly used to validate results, provided that the
sample size is sufficient. The idea of cross validation is to address the question of
whether the hypothesised model from one sample is replicated in a second
independent sample from the same population, so that the cross validation of
estimated parameters and relationships within and between the constructs can be
measured. The procedure can also be used to see whether the estimated parameters
and relationships within the constructs are meaningful and equal. Since the sample
size for this study (n = 529) is sufficient to split into two sub-samples to meet the
basic sample size requirement for SEM, this study used the split sample method for
validating the results (Anderson and Gerbing 1988).
5 For the purposes of this thesis, the argument regarding the one-step versus the two-step approach was kept to a minimum. A more detailed analysis can be found in Anderson and Gerbing’s (1988) work.
Chapter Eight – Multivariate Data Analysis
-213-
calibration sample, and the second sample will be used for validation (Anderson and
Gerbing 1988).
The two split samples were randomly chosen from the entire sample. Using
this method, the overall measurement model was tested to see whether the results
from the validation sample replicated the results from the calibration sample in terms
of the estimated parameters and relationships among the constructs. The final model
for the combined sample6
6 The only purpose for splitting the sample was to provide a measure of cross validation for the research.
, obtained via CFA, was tested to see whether the
hypothesised model fits the sample without any major changes in parameter
estimation (Anderson and Gerbing 1988).
Chapter Eight – Multivariate Data Analysis
-214-
8.3 Factor Analysis
This section outlines the choices made and the results associated with the
Exploratory Factor Analysis and Structural Equation Modelling methods.
8.3.1 Exploratory Factor Analysis
This thesis follows the scale development procedure proposed by Rossiter
(2002)7
7 See Chapter Five - Section 5.7.
. Using SPSS software, the study utilised Exploratory Factor Analysis (EFA)
during the measure purification step, to explore the underlying relationships between
the items that are hypothesised to measure the same construct. EFA was also
employed to ensure that all the hypothesised items for a construct emerged as separate
factors. This section explores the principles used to perform EFA, along with the main
results.
8.3.1.1Factor Extraction
Factor extraction in EFA only makes conceptual sense when the correlations
between individual items are high enough to justify the reduction of data to a smaller
number of factors (Hair et al. 2006). Two tests are commonly used to assess the
appropriateness of factor analysis: the Kaiser-Meyer-Olkin measure of sampling
adequacy (KMO), which is an index that reaches one when all variables are totally
predicted by other variables; and Bartlett’s test of sphericity, which tests for the
presence of correlations between variables. It is generally considered that factor
extraction is possible for KMO values of .7 and above, and when Bartlett’s test of
sphericity is significant, i.e. the significance value is .05 or higher (Hair et al. 1998).
The KMO value and the significance of Bartlett’s test of sphericity for this study are
reported below (see Table 8.3).
Chapter Eight – Multivariate Data Analysis
-215-
Table 8.3: KMO and Bartlett’s test of Sphericity Kaiser-Meyer-Olkin Measure of
Sampling Adequacy .733
Bartlett’s Test of Sphericity
Chi-Square 561.43
Degrees of Freedom 630
Significance .000
Cumulative covariance 60.72%
Once the suitability of factor extraction had been ascertained, the next decision
concerned the number of factors to extract. There are three main ‘rules’ used in
marketing to ascertain how many factors to extract; these rules are:
1. Number of Eigenvalues above 1, which is also known as the Kaiser criterion
method;
2. Scree plot sudden drop; and,
3. Horn’s parallel analysis, where only the Eigenvalues obtained from the
analysis which are higher than Eigenvalues produced by a set of random data
are retained.
The three rules rarely concur, but recent simulation studies tend to suggest that
Horn’s parallel analysis is the most suitable method for determining the correct
number of components because it yields a better factor selection (Horn 1965; Zwick
and Velicer 1986). While Horn’s parallel analysis results were given priority, all three
criteria mentioned above were reviewed, and where they produced no clear
consensus, analyses with a different number of factors to extract were also tested, and
the most interpretable solution was retained.
With regard to factor rotation, oblique (or oblimin) rotation was selected in
preference to orthogonal rotation, since the dimensions of the constructs were
expected to be correlated (Hair et al. 2006). To facilitate interpretation, only loadings
greater than .45 were requested and displayed in all rotation results.
Chapter Eight – Multivariate Data Analysis
-216-
The results of the EFA analysis indicate that the EFA test is valid, with a
KMO score of .77 and a significance of level of .000. The factor extraction
demonstrates an emergence of nine factors with a total covariance of 60.72% (see
Table 8.4). The results confirm all the hypotheses proposed in the main model.
Chapter Eight – Multivariate Data Analysis
-217-
Table 8.4: Rotated Component Matrix
Item Construct F1 F2 F3 F4 F5 F6 F7 F8 F9 Show fairness in transactions
(IN) .70
Always achieves consistency in service delivery
.71
Always keeps its word .82 The ….. employees treat me with
respect .71
Is open to my needs (BN)
.77 Acts in a caring manner .74 Keeps its customers' best interest
in mind during most transactions
.81
Is receptive to my needs .70 Has adequate skills to deliver the
right service
(CP) .79
Is efficient .70 The ….. employees can
competently handle most of my requests
.82
The ….. employees can be relied upon to know what they are doing
.74
Shares the same values as me (VA)
.82 Is ethical while dealing with me .73 Is interested in more than just
making profit out of customers .87
The ….. employees act as if they value the customer .71
Has stability in its operations (CO)
.74 Is always reliable .71 Always matches customers'
expectations .81
Always satisfies customers' needs
.78
Communicates clearly (CM)
.80 Respond immediately when
contacted .74
Inform me immediately of any problems
.80
Always communicates with me .76 I will transact with this ….. again
for future visits
(BL)
.85
I will try new services that are provided by this …..
.73
I will recommend other people to visit to this …..
.86
I will say positive things to other people about the services provided at this …..
.77
I will continue to visit this ….. even if the service charges are increased moderately
(AL)
.79
I have strong preference to this …..
.72
I will keep visiting this ….. regardless of everything being changed somewhat
.79
I am likely to pay a little bit more for using the services of this …..
.74
The ….. employees are professional and dedicated to customers
(TW)
.73
The ….. respond caringly when I share my problems
.75
My ….. is always honest with me .82 Overall I feel I can trust my ….. .70
F = Factor number; Integrity (IN), Benevolence (BN), Competence (CP), Value Alignment (VA), Consistency (CO), Communication (CM), Behavioural Loyalty (BL), Attitudinal Loyalty (AL),
Trustworthiness (TW)
Chapter Eight – Multivariate Data Analysis
-218-
8.3.2 Confirmatory Factor Analysis
Confirmatory Factor Analysis (CFA) has been utilised in this study to confirm
that the items load onto their respective constructs. The CFA technique is employed to
obtain the amount of measurement error in the model, to validate the multi-factorial
model, and to determine group effects on the factors (Jöreskog 1993).
Initially the overall model sample was used for the CFA analysis (see Table
8.88
8 The model was split into two parts for the purposes of clarification; it is easier to illustrate the model in two parts rather than as one single model. However, it has been analysed as one complete model using the SEM procedure.
, page: 223). For the input file, after specifying four observed indicators, the
variance/covariance matrix was assigned to a symmetric matrix with all parameters
free to be estimated. Finally, fitted variance/covariance matrix errors were tested.
The results of the CFA were acceptable since there was a Chi-square value of
635.830 with 570 degrees of freedom (p < .001). According to Wheaton et al. (1977),
a ratio of approximately five or less is ‘beginning to be reasonable’. In the current
study, however, degrees of freedom ratios in the range of 2:1 or 3:1 are indicative of
an acceptable fit between the hypothetical model and the sample data. Carmines and
McIver (1998:80) argue that “...different researchers have recommended using ratios
as low as 2 or as high as 5 to indicate a reasonable fit”. Byrne (1998:55) proposed that
a ratio greater than 2 represents an inadequate fit. As can be noted from Table 8.5, the
(χ2/ df) provides a ratio of 1.11, which is an acceptable ratio according to all the
researchers mentioned above.
As shown in Table 8.5, the factor loadings of items are all high (above 0.7)
and are significant for all constructs. The standard errors of variance associated with
the factor loadings are all significant (p<0.001), with values of p ranging from 0.022
to 0.0517. In other words, the convergent validity of the constructs is confirmed
(further discussion of these results will follow later in this chapter).
Chapter Eight – Multivariate Data Analysis
-219-
Table 8.5: Factor Loading and Standard Error for the Constructs
Factor Factor Estimate Standard Error
Trustworthiness Consistency .7 .048 Trustworthiness Competence .75 .022 Trustworthiness Integrity .89 .062 Trustworthiness Benevolence .91 .039 Trustworthiness Value alignment .75 .023 Trustworthiness Communication .74 .049
Behavioural loyalty Trustworthiness .82 .033 Attitudinal loyalty Trustworthiness .94 .043 Attitudinal loyalty Behavioural loyalty -.035 .051
According to Hair et al. (2006), high correlations between latent constructs
indicate that discriminant validity of constructs may exist in the sample. In the case of
this research, there was no high correlation between constructs, as shown in Table 8.6
below. The correlation between constructs ranged between -.066 and .173, which is an
acceptable level of correlation between the constructs (Tabachnick and Fidell 2007).
Table 8.6: Correlations between the Constructs
Construct Construct Estimate Communication Value alignment .055 Communication Benevolence .033
Integrity Communication .079 Competence Communication -.004 Consistency Communication .050
Value alignment Benevolence -.001 Integrity Value alignment -.033
Competence Value alignment -.057 Consistency Value alignment .173
Integrity Benevolence -.016 Competence Benevolence .158 Consistency Benevolence -.002
Integrity Competence .015 Consistency Integrity .050 Consistency Competence -.066
Because the viability of individual estimated values should be determined at
an initial stage when assessing the fit of individual parameters in a model, the
estimated parameters were examined in terms of not only their correct sign and size,
but also with regard to their consistency with the underlying theory. Unreasonable
Chapter Eight – Multivariate Data Analysis
-220-
estimates (with correlation values greater than one) were not found in the CFA results
for the measurement model (Tabachnick and Fidell 2007).
As the second step in the assessment of parameters (shown in Table 8.7
below), the squared multiple correlations (R2) were examined to see whether the
measurement model properly represented the observed items (Byrne 2001;
Tabachnick and Fidell 2007). These correlations were also assessed to determine the
indicator and construct reliability. As presented in Table 8.7, the squared multiple
correlations ranged from .31 to .67, which according to Nagelkerke (1991) the value
of R2 should be between 0 and 1, with 0 denoting that model does not explain any
variation and 1 denoting that it completely explains the observed variation, since
achieving a perfect fit is impossible (Nagelkerke 1991) the results of R2 are acceptable
to explain the model.
The CFA results indicate a Root Mean Square Error of Approximation
(RMSEA) of .015. RMSEA explains that the error of approximation in the population;
values should be less than .05 for a good fit (Hair et al. 2006; Tabachnick and Fidell
2007). Similarly, other fit indices also indicated a good fit, which suggested that the
estimate parameters need not be modified (see section 8.3.3 for a full discussion on fit
indices). These results also indicated that the initial hypothesised model was reliable
and valid (Hair et al. 2006). Therefore, it was decided that no modification indices
need be employed to improve the general fit of the results and the overall
measurement model (Hair et al. 2006).
Composite reliability has been calculated for each construct included in the
proposed model. This form of reliability in similar to Cronbach’s Alpha9
9 See Chapter Seven - Section 7.6.1, for a full discussion on Cronbach alpha reliability test.
, it reflects
the internal consistency of the items measuring a particular construct (Fornell and
Larcker 1981). Both the composite reliability and the variance extracted estimates are
shown in Table 8.7. Fornell and Larcker (1981) recommend a minimum composite
reliability of .60. An assessment of the composite reliabilities showed that all
constructs meet that minimum acceptable level advocated by (Fornell and Larcker
1981).
Chapter Eight – Multivariate Data Analysis
-221-
The variance extracted assesses the amount of variance that is explained by an
underlying factor in relation to the amount of variance due to measurement error. For
instance, the variance estimate for integrity was 0.62, meaning that 62% of the
variance is explained by the integrity construct, and 38% is subjected to measurement
error. Fornell and Larcker (1981) argue that constructs should obtain estimates of .50
or larger. Estimates lower than .50 signify that variance due to measurement error is
larger than the variance extracted by the construct. The variance extracted in Table
8.7 estimates all meet this minimum threshold; however, value alignment extracted
variance was to some extent lower than the rest of the items, because it still met the
required threshold level, it was decided not to take any action to improve the
construct.
Based on the above argument, the validity of the latent construct as well as the
related constructs is satisfactory. It should also be mentioned that Tabachnick and
Fidell (2007), noted that the variances extracted estimate test is conservative;
reliabilities can be acceptable even if variances extracted estimates are less than .50.
Chapter Eight – Multivariate Data Analysis
-222-
Table 8.7: Results of CFA for the Overall Measurement Model
Constructs Items Construct Reliability
Composite Reliability
10
Extracted Variance
Factor Loading
Squared Multiple
Correlations Integrity (IN) .73 .66 .62 IN1 .59 .32 IN2 .81 .33 IN3 .57 .67 IN4 .56 .34 Benevolence (BN) .76 .67 .62 BN1 .56 .49 BN2 .78 .40 BN3 .63 .60 BN4 .70 .31 Competence (CP) .77 .67 .61 CP1 .64 .51 CP2 .78 .35 CP3 .59 .61 CP4 .71 .41 Value Alignment (VA) .80 .68 .59 VA1 .57 .60 VA2 .85 .41 VA3 .64 .73 VA4 .77 .33 Consistency (CO) .76 .66 .72 CO1 .63 .40 CO2 .57 .32 CO3 .77 .59 CO4 .71 .51 Communication (CM) .789 .67 .64 CM1 .65 .56 CM2 .75 .38 CM3 .61 .57 CM4 .74 .42 Behavioural Loyalty (BL) .82 .68 BL1 .67 .67 BL2 .83 .38 BL3 .61 .70 BL4 .82 .45 Attitudinal Loyalty (AL) .76 .66 AL1 .71 .50 AL2 .60 .37 AL3 .71 .51 AL4 .63 .40 Trustworthiness (TW) .75 .66 TW1 .57 .33 TW2 .67 .45 TW3 .81 .66 TW4 .57 .32
10 See appendix Twelve for the full calculation of the composite reliability tests.
Chapter Eight – Multivariate Data Analysis
-223-
Next, the factor loadings were evaluated and found to fall within a range
between .83 and .56. This calculation was employed to determine the relative
importance of the observed variables as indicators of the constructs (Kline 1998).
Lastly, the extracted variances that represent the overall amount of variance in the
indicators accounted for by the latent constructs and values were calculated. For each
construct, the values were: IN (.52); BN (.62); CP (.51); VA (.56); CO (.72); and CM
(.64). These values all exceed the recommended level of .50 (Hair et al. 2006).
8.3.3 Goodness of Fit Indices Discussion
After completing the estimation the parameters of the measurement model, the
hypothesised overall model was examined using three types of fit index: absolute fit
indices; incremental fit indices; and parsimonious fit indices (Tabachnick and Fidell
2007). The results of the goodness-of-fit statistics were reported in Table 8.8.
Table 8.8: Overall Model Fit Indices Overall Model Fit Goodness-of-Fit Indices
Absolute Fit Measures
Chi-square (χ2) of the estimated model χ2= 635.830 DF= 570
Goodness-of-fit index (GFI) .938
Root mean square error of approximation
(RMSEA)
.015
Root mean square residual (RMR) .021
Incremental Fit Measures
Adjusted goodness-of-fit index (AGFI) .927
Normed fit index (NFI) .91
Parsimonious Fit Measures
Comparative fit index (CFI) .987
Relative fit index (RFI) .878
Parsimony normed fit index (PNFI) .805
Parsimony goodness-of-fit index (PGFI) .803
Incremental fit index (IFI) .987
Chapter Eight – Multivariate Data Analysis
-224-
First, the absolute fit index directly measures how well the a priori model
reproduces the collected sample data. In other words, it is applied to measure how
closely the model matches a perfect fit (Tabachnick and Fidell 2007; Maruyama
1997). These indices include: Chi-square (χ2) of the estimated model; goodness-of-fit
index (GFI); root mean square residual (RMR); and root mean square error of
approximation (RMSEA) (Byrne 2001; Kline 1998; Hoyle 1995).
The Chi-square (χ2) of the estimated model was examined to test the closeness
of fit between the unrestricted sample covariance matrix and the restricted covariance
matrix. The χ2 value of 635.830 with 570 degrees of freedom was not statistically
significant with p = .029, thereby suggesting that the hypothesised overall
measurement model with nine constructs and thirty-six indicators was appropriate and
should be accepted at this statistical level.
The goodness-of-fit index (GFI) that was used to compare the hypothesised
model with null model yielded a value of .938. This index takes a value from zero to
1.00, with values to close to 1.00 being indicative of a good fit (Byrne 2001).
Accordingly, the result of the GFI for this study produced an acceptable level of fit.
The value of the root mean square residual (RMR) was .021. Since this index
is used to estimate the average residual value derived from the fitting of the variance
covariance matrix for the hypothesised model (nine constructs with thirty-six
indicators) to the variance-covariance matrix of the sample data, a smaller value
indicates a better fitting model. This value of .021 represents the average value across
all standardised residuals, ranging from zero to 1.00. For a well-fitting model, this
value should be less than .05. The RMR value of .021 for the current study
represented the correlations to within an average error of .021, which was acceptable
as a well fitting model for this study (Tabachnick and Fidell 2007).
Next, the root mean square error of approximation (RMSEA) is an index used
to quantify model misfit; a value of less than .05 indicates a good model fit, and
values higher than .08 indicate sensible errors of approximation in the population
(Hair et al. 2006). The value of RMSEA for the hypothesised measurement model
was .015, which falls inside the acceptable level. Furthermore, this value also
conceded a 90% confidence interval, ranging from .00 to .037, and the p-value for the
Chapter Eight – Multivariate Data Analysis
-225-
test of closeness-of-fit equalled 1.00. Consequently, the value of RMSEA of .015,
falling within the bounds of .00 and .037, represented an excellent degree of
precision.
Generally, based on the examination of the absolute fit statistical indices, it
can be concluded that the hypothesised measurement model constituted a well-fitting
model for the collected data. Consequently, further analysis such as the execution of
the measurement model was possible and proved valid (Anderson and Gerbing 1988;
Tabachnick and Fidell 2007).
The incremental fit indices were examined as part of goodness of fit statistics,
(Maruyama 1997). These tests were used to assess the proportional improvement in fit
achieved by comparing the target model with a more restricted, nested baseline model
(Maruyama 1997). These indices include the adjusted goodness-of-fit index (AGFI),
the non-normed fit index (NNFI), and the normed fit index (NFI) (Kline 1998).
The adjusted goodness-of-fit index (AGFI) may also be conceived as an
absolute goodness of fit index (Byrne 2001). This index is similar to the GFI, but is
marginally different in that it is adjusted for the number of degrees of freedom in the
specified model (Hair et al. 2006). Given that the value of the AGFI was .927, which
exceeded the recommended level of .90, the hypothesised model was shown to fit
reasonably well.
The NFI represents the proportion of the total covariance among the observed
variables that is explained by a target model, using the null model as a baseline (Hair
et al. 2006). The possible values of NFI range from zero to 1.00, with a value of
greater than .95 being acceptable to indicate a well fitting model. For this study the
value of NFI was .91, suggesting that the model fit the data fairly well (Hair et al.
2006). Overall, the hypothesised model demonstrated a healthy fit to the data.
Finally, the parsimonious fit indices represent information about an evaluation
between models of differing complexity and objectives, by evaluating the fit of the
model versus the number of estimated coefficients needed to achieve that level of fit
(Tabachnick and Fidell 2007). These indices comprise the parsimony goodness-of-fit
Chapter Eight – Multivariate Data Analysis
-226-
index (PGFI), the parsimony normed fit index (PNFI), the comparative fit index
(CFI), the incremental fit index (IFI), and the relative fit index (RFI).
The parsimony goodness of fit index (PGFI) relates to the parsimony of the
model, and takes into account the complexity of the hypothesised model in the
assessment of overall model fit (Tabachnick and Fidell 2007). The values vary
between zero and 1.00, with higher values indicating greater model parsimony
(Tabachnick and Fidell 2007). As shown in Table 8.5 above, the value of the PGFI
was .803, suggesting that the hypothesised model fit the data parsimoniously. The
parsimony normed fit index (PNFI) explains the complexity of the model in its
assessment of goodness-of-fit. Essentially this index is used for the comparison of
models with differing degrees of freedom. A higher value of the PNFI indicates a
better model fit. The value of the PNFI in this analysis was .805, which is an
acceptable value for a well fitting model.
The incremental fit index (IFI) represents the subjects of parsimony and
sample size associated with NFI, and is used to compare a restricted model with a full
model using a baseline null model (Byrne 2001). The value of the comparative fit
index (CFI) measures the improvement in non-centrality by going from the least
restrictive model to the most saturated model. The values of the CFI range from zero
to 1.00. The relative fit index (RFI) is equivalent to CFI. Higher values of IFI, CFI,
and RFI all indicate a better model fit to the data. As shown in Table 8.5 above, the
values of IFI, CFI, and RFI were .987, .987 and .878 respectively, signifying that all
these values were sufficient to indicate a well fitting model (Byrne 2001).
As a result of the above analyses, the review of the three types of goodness of
fit indices for the overall measurement model revealed consistent patterns of values of
fit indices, and showed that the model was well fitted to the observed data. This
indicates that the measurement model is reliable and valid in representing the sample.
In addition to these multiple criteria, the examination of the theoretical and practical
aspects of the measurement model with nine constructs and thirty-six observed
indicators supported the assessment that this measurement model was adequate in
describing the collected data.
Chapter Eight – Multivariate Data Analysis
-227-
8.3.4 CFA with Calibration and Validation Sample
The evaluation of goodness-of-fit statistics for the calibration and validation
sample was done by three types of fit indices for the entire measurement model;
absolute fit indices, incremental fit indices, and parsimonious fit indices (Hair et al.
2006). As shown in Table 8.9 below, all the fit indices yielded acceptable levels to
indicate a well-fitting model to the data.
The absolute fit index, which directly measures how well a priori model
reproduces the collected sample data, showed that the χ2 of the estimated model is
613.190 with 570 degrees of freedom for the calibration sample, while the Chi-square
for the validation sample was 625.425 with 570 degrees of freedom. The goodness of
fit index (GFI) was .96 for the calibration sample and .90 for the validation sample.
The root mean square residual (RMR) was .028 for the calibration sample and .031
for the validation sample, and the root mean square error of approximation (RMSEA)
was .017 for the calibration sample and .019 for the validation sample. The
incremental fit indices, which evaluate the proportional improvement in fit achieved
by comparing a target model with a more restricted, nested base line model, produced
a normed fit index (NFI) of .801 and .818 for the calibration and validation samples
respectively.
Table 8.9: Results of CFA for Calibration and Validation Split Samples
Overall model fit Goodness-of-Fit Statistics
(Calibration sample)
Goodness-of-Fit Statistics
(Validation sample)
Absolute Fit Measures
Chi-square (χ2) of the estimated model χ2 = 613.190
DF = 570
χ2 = 625.425
DF = 570
Goodness-of-fit index (GFI) .93 .90
Root mean square error of
approximation
(RMSEA)
.017 .019
Root mean square residual (RMR) .028 .031
Incremental Fit Measures
Adjusted goodness-of-fit index (AGFI) .872 .865
Normed fit index (NFI) .801 .818
Parsimonious Fit Measures
Chapter Eight – Multivariate Data Analysis
-228-
Comparative fit index (CFI) .982 .980
Relative Fit Index (RFI) .780 .798
Parsimony normed fit index (PNFI) .725 .740
Parsimony goodness-of-fit index(PGFI) .762 .757
Incremental fit index (IFI) .983 .981
Finally, the parsimonious fit indices provided information for comparison
between models of differing complexity and objectives, by evaluating the fit of the
model versus the number of estimated coefficients needed to achieve that level of fit.
The parsimony goodness-of-fit index (PGFI) was .762 for the calibration sample and
.757 for the validation sample; the parsimony normed fit index (PNFI) was.725 and
.740 for the calibration and validation samples respectively, and the comparative fit
index (CFI) was .982 and .980 for the calibration sample and the validation sample.
This review of the three types fit index comparing both samples reveals
consistent patterns in the values of the fit indices, supporting the conclusion that the
model fits the data well.
Despite these reliable results from the split sample, this study included a
further analysis based on the entire sample (n=529) because the overall score in the
goodness of fit indices were to some extent better with regard to the split samples than
with the entire sample. Accordingly, additional analysis such as full structural
equation modelling for the hypotheses tests was possible.
8.3.5 CFA analysis for the Model’s Constructs
In this section an in-depth discussion concerning each of the model’s nine
constructs is presented individually, to enable insights to be gained into which item
has the most influence on that particular construct in terms of the regression weights
(factor loading). Analysis of the regression weights related to the paths from the latent
variables to their respective indicator variables will increase our understanding about
which item(s) have a higher influence, depending on their factor loading, on the
respective construct (the paths in the measurement model). This will help to generate
an understanding of how to control the model, enhancing or reducing the effects of
different constructs.
Chapter Eight – Multivariate Data Analysis
-229-
8.3.5.1Integrity
Integrity was measured in the model with four items (see Table 8.10 below).
The respondents were asked to indicate how much each of the items reflected the
level of integrity they attributed to their respective service provider. As can be seen
from the factor loading of each item, all the items had a positive impact on integrity,
which indicates that all the items were significant and related to the construct of
integrity.
Table 8.10: Integrity Regression Weights
Integrity (IN) items Regression weights
IN1: Show fairness in transactions .56
IN2: Always achieves consistency in service delivery .57
IN3: Always keeps its word .8111
IN4: The ….. employees treat me with respect
.59
Figure 8.5: Integrity Regression Weights
The regression weights, in Figure 8.5, ranged from .56 to .81, in which item
IN1 (‘Show fairness in transactions’) was found to be the least effective item among
the observed indicators for integrity. In comparison, item IN3 (‘Always keeps its
word’) was the most important item, having a loading of .81. The construct reliability
11 The highest values are highlighted.
Integrity IN3
IN4
IN2
IN1
E11
E12
E10
E9
1
.33
.30
.15
1
.56
.30
.57
.81
1
1
.59
.16
Chapter Eight – Multivariate Data Analysis
-230-
for integrity is .73, which exceeds the recommended level of .70 (Hair et al. 2006).
Overall the four indicators for integrity were sufficient to represent the construct.
8.3.5.2 Competence
Competence was measured in the hypothesised model with four items (see
Table 8.11 below). The respondents were asked to express to what extent each of the
items reflected the level of competence they expected from their respective service
provider. As can be noted from the regression weights related to each item, all the
items have a positive effect on competence, which confirms that all the items were
significant and related to the construct.
Table 8.11: Competence Regression Weights
Competence (CP) items Regression weights
CP1: Has adequate skills to deliver the right service .71
CP2: Is efficient .59
CP3: The ….. employees can competently handle most of my
requests
.78
CP4: The ….. employees can be relied upon to know what they are
doing
.64
Figure 8.6: Competence Regression Weights
The regression weights, in Figure 8.6, ranged from .59 to .78, wherein item
CP2 (‘is efficient’) was found to be the least effective item among the observed items
for competence. In comparison, item CP3 (‘the ….. employees can competently
Competence CP3
CP4
CP2
CP1
E7 E8
E6
E5
1
.26
.37
.23
1
.71
.33
.59
.78
1
1
.64
.23
Chapter Eight – Multivariate Data Analysis
-231-
handle most of my requests’) was the most significant item, having a loading of .78.
The construct reliability for competence is .77, which exceeds the recommended level
of .7 (Hair et al. 2006). Generally, the four indicators for competence were sufficient
to represent the construct.
8.3.5.3 Value Alignment
Value alignment was evaluated in the model by four items (see Table 8.12
below). The respondents were asked to indicate to what extent each of the items
reflected the perceived level to which their values aligned with their respective service
provider. As can be seen from the regression weights for each item, all the items had a
positive impact on value alignment, which suggests that all the items were important
and related to the construct of value alignment.
Table 8.12: Value alignment Regression Weights
Value Alignment (VA) items Regression weights
VA1: Shares the same values as me .77
VA2: Is ethical while dealing with me .64
VA3: Is interested in more than just making profit out of customers .85
VA4: The ….. employees act as if they value the customer .57
Figure 8.7: Value alignment Regression Weights
The regression weights, in Figure 8.7, ranged from .57 to .85. Item VA4 (‘The
….. employees act as if they value the customer’) was found to be the least effective
item among the observed indicators for value alignment. In comparison, item VA3
Value Alignment
VA3
VA4
VA2
VA1
E15
E16
E14
E13
1
.24
.40
.18
1
.77
.43
.64
.85
1
1
.57
.21
Chapter Eight – Multivariate Data Analysis
-232-
(‘Is interested in more than just making profit out of customers’) was the most
significant item, having a loading of .85. The construct reliability for value alignment
is .8, which exceeds the recommended level of .7 (Hair et al. 2006). Overall, the
specified four indicators for value alignment were sufficient to represent the construct.
8.3.5.4 Communication
Communication was measured in the model by four items (see Table 8.13
below). The respondents were asked to indicate how well each of the items reflected
the level of communication from their respective service provider. As can be seen
from the factor loading for each item, all the items had a positive correlation with
communication, which signifies that all the items were related to the construct of
communication.
Table 8.13: Communication Regression Weights
Communication (CM) items Regression weights
CM1: Communicates clearly .74
CM2: The ….. respond immediately when contacted .61
CM3: Inform me immediately of any problems .75
CM4: The ….. always communicates with me .65
Figure 8.8: Communication Regression Weights
The regression weights, in Figure 8.8, ranged from .61 to .75, with item CM4
(‘The ….. always communicates with me’) was found to be the least effective item
among the observed indicators for communication. In comparison, item CM3
Communication CM3
CM4
CM2
CM1
E19
E20
E18
E17
1
.27
.31
.25
1
.74
.34
.61
.75
1
1
.65
.25
Chapter Eight – Multivariate Data Analysis
-233-
(‘Inform me immediately of any problems’) was the most important item, with a
loading of .75. The construct reliability for communication is .78, which exceeds the
recommended level of .7 (Hair et al. 2006). Overall the four indicators for
communication were sufficient to represent the construct.
8.3.5.5 Benevolence
Benevolence was measured in the model by four items (see Table 8.14 below).
The respondents were asked to indicate how well each of the items reflected the level
of benevolence they perceived in their respective service provider. As can be seen
from the factor loading for each item, all the items had a positive impact on
benevolence, which indicates that all the items were significant and related to the
construct of benevolence.
Table 8.14: Benevolence Regression Weights
Benevolence (BN) items Regression weights
BN1: Is open to my needs .7
BN2: Acts in a caring manner .63
BN3: Keeps its customers’ best interest in mind during most
transactions
.78
BN4: Is receptive to my needs .56
Figure 8.9: Benevolence Regression Weights
The regression weights, in Figure 8.9, ranged from .56 to .78. Item BN4 (‘Is
receptive to my needs’) was the least effective, with a factor loading of .563. In
Benevolence BL3
BL4
BL2
BL1
E23
E24
E22
E21
1
.27
.34
.24
1
.7
.39
.63
.78
1
1
.56
.18
Chapter Eight – Multivariate Data Analysis
-234-
comparison, item BN3 (‘Keeps its customers’ best interest in mind during most
transactions’) was the most effective item, with a factor loading of .78. The construct
reliability for benevolence is .73, which exceeds the recommended level of .7 (Hair et
al. 2006). Generally, the four items for benevolence were sufficient to represent the
construct.
8.3.5.6 Consistency
Consistency was measured in the model by four items (see Table 8.15 below).
The respondents were asked to indicate how well each of the items reflected the level
of consistency they expected from their respective service provider. As can be seen
from the factor loading of each item, all the items had a positive impact on
consistency, which indicates that all the items were significant and related to the
construct of consistency.
Table 8.15: Consistency Regression Weights
Consistency (CO) items Regression weights
CO1: Has stability in its operations .63
CO2: Is always reliable .57
CO3: Always matches customers’ expectations .77
CO4: Always satisfies customers’ needs .71
Figure 8.10: Consistency Regression Weights
The regression weights, in Figure 8.10, ranged from .57 to .77. Item CO2 (‘Is
always reliable’) was found to be the least effective item among the observed
Consistency CO3
CO4
CO2
CO1
E3
E4
E2
E1
1
.41
.43
.28
1
.63
.32
.57
.77
1
1
.71
.28
Chapter Eight – Multivariate Data Analysis
-235-
indicators for consistency, with a factor loading of .57. In comparison, item CO3
(‘Always matches customers’ expectations’) was the most important item, having a
loading of .77. The construct reliability for consistency is .76, which exceeds the
recommended level of .7 (Hair et al. 2006). Overall the specified four indicators for
consistency were sufficient to represent the construct.
8.3.5.7 Trustworthiness and Loyalty
Trustworthiness and loyalty were measured in the model by four items for
each construct (see Table 8.16 below). The respondents were asked to indicate how
well each of the items reflected the level of trustworthiness and loyalty they received
from their respective service provider. As can be seen from the regression weights for
each item, all the items had a positive impact on their constructs, which indicates that
all the items were significant and related to the construct.
Table 8.16: Trustworthiness Regression Weights
Trustworthiness (TW) items Regression weights
TW1: The ….. employees are professional and dedicated to
customers
.57
TW2: The ….. respond caringly when I share my problems .67
TW3: My ….. is always honest with me .81
TW4: Overall I feel I can trust my ….. .57
Figure 8.11: Trustworthiness, attitudinal and behavioural Loyalty Regression Weights
Behavioural loyalty
Attitudinal loyalty
Trustworthiness AL4
AL3
AL2
AL1
BL1
BL2 TW4
TW3
TW2
BL3
BL4
TW1
E28
E27
E26
E25
E29
E30
E31
E32
E33
E34
E36
E35
.25
.33
.18
.33
.18
.27 .40
.33
.22
.38
1
1
1
1
.57
.572
.81
.67
.71
.609
.71
.63
.82
.67
.617
.839
E37
E38
E39
.20
.28
.25
Chapter Eight – Multivariate Data Analysis
-236-
The regression weights, in Figure 8.11, for trustworthiness ranged from .57 to
.81. Item TW4 (‘Overall I feel I can trust my …..’) was found to be the least effective
among the observed items for trustworthiness, with a factor loading of .57. In
comparison, item TW3 (‘My ….. is always honest with me’) was the most important
item, with a loading of .81. The construct reliability for trustworthiness is .75, which
exceeds the recommended level of .7 (Hair et al. 2006).
Table 8.17: Behavioural Loyalty Regression Weights
Behavioural Loyalty (BL) items Regression weights
BL1: I will transact with this ….. again for future visits .82
BL2: I will try new services that are provided by this ….. .61
BL3: I will recommend other people to visit to this ….. .83
BL4: I will say positive things to other people about the services
provided at this …..
.67
Behavioural loyalty regression weights ranged from .61 to .83 (see Table 8.17
above). Item BL2 (‘I will try new services that are provided by this …..’) was found
to be the least effective item among the observed indicators for behavioural loyalty,
with a factor loading of .61. in comparison, item BL3 (‘I will recommend other
people to visit to this …..’) was the most important item, having a loading of .83. The
construct reliability for behavioural loyalty is .82, which exceeded the recommended
level of .7 (Hair et al. 2006).
Table 8.18: Attitudinal Loyalty Regression Weights
Attitudinal Loyalty (AL) items Regression weights
AL1: I will continue to visit this ….. even if the service charges are
increased moderately
.71
AL2: I have strong loyalty to this ….. .6
AL3: I will keep visiting this ….. regardless of everything being
changed somewhat
.71
AL4: I am likely to pay a little bit more for using the services of
this …..
.63
Chapter Eight – Multivariate Data Analysis
-237-
Attitudinal loyalty regression weights ranged from .6 to .71 (see Table 8.18
above). Item AL2 (‘I have strong loyalty to this …..’) was found to be the least
effective item among the observed indicators for behavioural loyalty, with a factor
loading of .6. In comparison, item AL3 (‘I will keep visiting this ….. regardless of
everything being changed somewhat’) was the most important item, with a loading of
.71. The construct reliability for attitudinal loyalty is .76, which exceeds the
recommended level of .70 (Hair et al. 2006).
8.3.6 Assessing Construct Validity
Convergent and discriminant validity can both be considered as subcategories
of construct validity (Hair et al. 2006; Tabachnick and Fidell 2007). To achieve
construct validity, the empirical evidence must be consistent with the theoretical logic
about the concepts. Convergent validity refers to the confirmation of the measurement
of a construct by the use of multiple methods. Convergent validity is demonstrated by
highlighting that the different items tapping their respective latent variables correlate
with each other to an acceptable degree. Acceptable goodness of fit measures for a
model indicates convergent validity; it is also the case that the pattern coefficients
should be at least .70 for all indicators using a common rule of thumb (Kline 1998).
Discriminant validity is an indication of the extent to which measures have a
low correlation with other measures which are associated with dissimilar concepts. In
other words, discriminant validity relates to the distinctiveness of constructs. The
correlation between two scales for two distinct constructs should not be high if we are
to infer discriminant validity. In applying this concept, the items employed to measure
different constructs in models should provide different results.
Conversely in terms of convergent validity, if the indicators which are
intended to measure a common underlying factor have comparatively high loadings
on that factor, convergent validity is achieved (Tabachnick and Fidell 2007). These
high loadings mean that strong correlations on the underlying theoretical construct are
achieved, and the measurement scales can be said to be measuring what they are
intended to measure (Kline 1998).
Chapter Eight – Multivariate Data Analysis
-238-
In SEM analysis, the standardised factor loading can be examined in order to
evaluate convergent validity. As can be seen in Table 8.6, presented earlier in this
chapter, the estimated standardised factor loading coefficients for the stated
underlying constructs produced statistically significant results at the p = .05 level. The
factor loading for each observed indicator exceeded the recommended level of t
(+1.96). Therefore, it can be concluded that the measurement scale achieved
convergent validity of the constructs.
To examine discriminant validity, this study utilised the procedures suggested
by Anderson and Gerbing (1988) to be sure that the constructs are not measuring the
same concept or idea. Accordingly, taking one pair of constructs at a time, the
outcomes of the unconstrained CFA were compared with similar results for a
constrained model. The unconstrained CFA model was allowed to co-vary without
restraint, while the constrained model fixed the factor covariance to zero, meaning
that there was no correlation between two constructs (Anderson and Gerbing 1988).
In other words, discriminant validity can be examined by constraining the estimated
correlation parameter between each pair of constructs to zero (the fixed model).
In the constrained model, the correlation parameter of zero indicates that the
two constructs are uncorrelated. Different parameters in the unconstrained model (the
free model) indicate that the correlation between factors was estimated (Hoyle 1995).
A Chi-square test between the constrained model and the unconstrained model was
performed. A significant Chi-square difference between the models provided evidence
of discriminant validity between the pairs of the constructs being tested (Anderson
and Gerbing 1988).
Furthermore, a significantly lower Chi-square value for the unconstrained
model indicates that discriminant validity is achieved, and the goodness of fit
statistics is improved in the unconstrained model when discriminant validity is
achieved (Klein 1998).
Chapter Eight – Multivariate Data Analysis
-239-
8.4 The Structural Model Process
As discussed earlier, this study adopted SEM in order to test its hypotheses. In
particular, SEM has been applied to test the hypotheses concerning relationships
among the observed and the unobserved variables (Hair et al. 2006). For the most
part, in previous research SEM has been considered as a way of testing a specified
theory about relationships between theoretical constructs (Tabachnick and Fidell
2007).
The structural equation model is used to test a hypothetical model that
prescribes relationships between latent constructs and observed variables that are
indicators of those latent constructs. The relationships between the constructs can be
identified by providing path coefficients (parameter values) for each of the research
hypotheses. Each estimated path coefficient can be tested for its respective statistical
significance regarding the relationships between the hypotheses, including a
consideration of standard errors (Hair et al. 2006).
The primary objectives of this study were to develop a theoretical model to
measure trustworthiness. The proposed framework can be seen as establishing a
theoretical foundation for trustworthiness by identifying the antecedents of trust –
those factors that will ultimately lead to trustworthiness. Customer loyalty is proposed
in the model as an outcome of trustworthiness.
Chapter Eight – Multivariate Data Analysis
-240-
The theoretical framework (shown in Figure 8.12) is a static dyadic model that
assesses relationships at a given moment. The model places trustworthiness at the
focal point, with determinants of trustworthiness classified as low and high level
determinants having a direct relationship with the central concept. These determinants
are listed with no particular ranking (integrity, competence, consistency,
communications, value alignment and benevolence). The third part of the model
covers the outcome of the model, illustrated by two types of customer loyalty
(behavioural and attitudinal loyalty), with the prediction that behavioural loyalty can
lead to attitudinal loyalty. The underlying theoretical importance of each construct has
been discussed in a previous chapter in this thesis (see Chapter Four section 4.6). The
results of the path analysis and structural model fitting will be discussed in the
following section.
Benevolence
Competence
Integrity
Consistency
Value alignment
Communication
Behavioural loyalty
Attitudinal loyalty
Trustworthiness
Low level
High level
H1
H8
H7
H9 H4
H5
H6
H3
H2
Figure 8.12: Complete Path Diagram with the Hypothesised Measurement Specifications
Chapter Eight – Multivariate Data Analysis
-241-
8.4.1 Path Analysis
Path analysis is an extension of the regression model, applied to test the fit of
the correlation matrix against two or more causal models that are being compared by
the researcher (Hoyle 1995). A path model is a diagram relating independent,
intermediary, and dependent variables. Single arrows indicate causation between
exogenous or intermediary variables and the dependent(s) (Hair et al. 2006). Arrows
also link the error terms with their respective endogenous variables. Double arrows
indicate correlations between pairs of exogenous variables (see Figure 8.5 above). A
regression analysis is carried out for pairs of constructs where the theoretical model
indicates causative relationships. The regression weights predicted by the model are
compared with the observed correlation matrix for the variables, and a goodness of fit
statistic is calculated.
Path analysis requires the standard assumptions of regression to be met. It is
particularly sensitive to model specification, because failure to include relevant causal
variables or the inclusion of extraneous variables can substantially affect the path
coefficients, which are used to measure the relative importance of various direct and
indirect causal paths among the dependent constructs. Such interpretations should be
performed in the context of comparing alternative models after measuring their
goodness of fit, as discussed in the section on SEM (see section 8.5 above). When the
variables in the model are latent variables measured by multiple observed indicators,
path analysis is termed SEM, and is treated separately (Tabachnick and Fidell 2007).
Exogenous variables in a path model are those with no explicit causes (no
arrows going to them, other than the measurement error term). If exogenous variables
are correlated, this is indicated by a double-headed arrow connecting them. Therefore,
endogenous variables are those that have incoming arrows (Tabachnick and Fidell
2007). Endogenous variables include intervening causal variables and dependents.
Intervening endogenous variables have both incoming and outgoing causal arrows in
the path diagram. The dependent variable(s) have only incoming arrows (Tabachnick
and Fidell 2007).
Chapter Eight – Multivariate Data Analysis
-242-
According to the standardised path coefficients presented in Figure 8.13, this
model is specified by the following path equations. Trustworthiness is influenced by
its antecedents, consistency, competence, benevolence, integrity, value alignment and
communication, by loadings of .10, .60, .30, .25, .60 and .50 respectively.
Furthermore, trustworthiness influences both behavioural and attitudinal loyalty by
.40 and .50 respectively. All the constructs were found to be significant (0 >. 05).
Through the path coefficients, it can be seen that both value alignment and
integrity have more effect on the creation of trustworthiness compared with the rest of
the variables. This finding has an impact on the study hypotheses in terms of
accepting or rejected the hypotheses, as will be discussed later in section 8.6.
It is generally acknowledged that most theoretical models are useful estimates
that do not fit perfectly in the population. In other words, the null hypothesis of a
perfect fit is not credible to begin with, and will, in the end, be accepted only if the
sample is not allowed to get too big. Jöreskog (1969: 200) noted that “such a
hypothesis [of perfect fit] may be quite unrealistic in most empirical work with test
Benevolence
Competence
Integrity
Consistency
Value alignment
Communication
Behavioural loyalty
Attitudinal loyalty
Trustworthiness
Low level
High level
.10
.4
.5
-.04 .25
.60
.50
.30
.60
Chi-square = 635.830 Degrees of freedom = 570 P = .029
Figure 8.13: The Path Loadings Diagram
Chapter Eight – Multivariate Data Analysis
-243-
data. If a sufficiently large sample were obtained this statistic would, no doubt,
indicate that any such non-trivial hypothesis is statistically untenable.” Furthermore,
“...in very large samples virtually all models that one might consider would have to be
rejected as statistically untenable. ... In effect, a non-significant chi-square value is
desired, and one attempts to infer the validity of the hypothesis of no difference
between model and data. Such logic is well known in various statistical research as
attempting to prove the null hypothesis” (Bentler and Bonett 1980: 591).
8.4.2 Testing Alternative Models
When applying SEM, it is recommended to test the model against alternative
models, to eliminate the possibility that the alternative models provide a better
explanation for the phenomena being studied. This also provides the researcher with
confirmation that the accepted model scores the best fit (Schumacker and Lomax
2004). The identification of each alternative model was based on theoretical grounds,
and all the relationships in the alternative models were clearly identified. Specifically,
four steps were followed for alternative model testing:
1. Generation of several theoretical models. These models have similar but
distinct relationships compared with the main proposed model of the study, are
based on the literature and the exploratory study surrounding the model’s
constructs.
2. Path model specifications, in which all relationships between the constructs
are clearly defined. This helped establishing the overall goodness of fit indices
when applying CFA, and will thereby indicate which model(s) provide a better
fit overall.
3. Best model fit: examined via SEM analysis and reporting the key goodness of
fit indices.
4. Final model evaluation and interpretation: the models’ reliability and validity
will be reported, as well as the significance levels for the path diagrams.
Several other models were developed, based on alternative path models and
theoretical rationale. These models are described below.
Chapter Eight – Multivariate Data Analysis
-244-
8.4.2.1 Alternative Model A
In the first alternative model A (see Figure 8.14 below), the model outcome
was consolidated into one construct labelled as ‘customer loyalty’. This is in
agreement with the study by Sideshmukh et al. (2002), which examined loyalty as one
construct regardless of several other studies that classified different levels of customer
loyalty (for example, Rundle-Thiele and Bennett 2001).
The rest of the constructs in model A remain the same as the main
hypothesised model in the study.
Benevolence
Competence
Integrity
Consistency
Value alignment
Communication
Customer Loyalty
Trustworthiness
Low level
High level
H1
H7
H4
H5
H6
H3
H2
Figure 8.14: Alternative Model A
Chapter Eight – Multivariate Data Analysis
-245-
8.4.2.2 Alternative Model B
Model B (see Figure 8.15 below) examines the low level antecedents of
trustworthiness. Model B is based on the argument that customers always seek the
cognitive level of the provided service (McAllister 1995); if the emotional level exists
it is an extra added value to the service, but it is not essential.
The two types of loyalty are retained as the model outcome, with a prediction
that behavioural loyalty can lead to the higher level attitudinal loyalty.
Competence
Consistency
Behavioural loyalty
Attitudinal loyalty
Trustworthiness
Low level
H1
H4
H3
H5 H2
Figure 8.15: Alternative Model B
Chapter Eight – Multivariate Data Analysis
-246-
8.4.2.3 Alternative Model C
The third alternative model C (see Figure 8.16 below) examines the opposite
situation, looking only at the high level antecedents of trustworthiness. This model is
based on the argument that customers always expect a cognitive level of
trustworthiness, but if they want to develop the relationship with the service provider,
several emotional factors has to exist in order to develop trustworthiness.
Once more, the two types of loyalty are retained as the model outcome, with a
prediction that behavioural loyalty can lead to the higher level attitudinal loyalty.
Benevolence
Integrity
Value alignment
Communication Behavioural
loyalty
Attitudinal loyalty
Trustworthiness
High level
H6
H5
H7
H2
H3 H4
H1
Figure 8.16: Alternative Model C
Chapter Eight – Multivariate Data Analysis
-247-
8.4.2.4 Alternative Model D
The final alternative model, model D (see Figure 8.17 below), examines
trustworthiness in terms of arguments presented by Mayer et al. (1995)12
The model outcome consists of two types of customer loyalty, with no
predicted relationship between the two types of customer loyalty.
. This model
assumes that trustworthiness can be conceptualised by the three main antecedents
proposed in Mayer et al.’s (1995) classic work, and used extensively in various other
studies on trust (although not trustworthiness). For instance, see Caldwell and
Clapham (2003).
12 See Figure 4.3.
Benevolence
Competence
Integrity
Behavioural loyalty
Attitudinal loyalty
Trustworthiness
H5
H4
H3
H2
H1
Figure 8.17: Alternative Model D
Chapter Eight – Multivariate Data Analysis
-248-
8.4.3 Evaluation of Fit for the Alternative Models
All the four models were analysed used AMOS software (Byrne 2001). The
Chi-squares of the four models compared with the main hypothesised model are
shown in Table 8.19, and the Chi-square differences between the models are
presented in Table 8.20. It is evident that the main proposed model has a much better
fit with the data than any of the alternative models. The main model is the only model
with fit indices within an acceptable range.
Therefore, it can be concluded that the main proposed model tested earlier in
this chapter is statistically the most acceptable and appropriate model. Moreover,
discussion and conclusions will be based upon this model. Criterion validity13
Model
was
also achieved when the instrument was compared with other models. In addition,
Table 8.19: Goodness of Fit Indices for the Alternative Models
χ2 DF RMSEA CFI PGFI GFI
Main (9 constructs) 635.830 570 .015 .987 .803 .938
A (8 constructs) 1426.329 572 .053 .833 .737 .858
B (5 constructs) 1254.07 511 .18 .81 .71 .821
C (7 constructs) 394.175 337 .008 .916 .788 .909
D (6 constructs) 1090.797 346 .064 .814 .731 .857
Table 8.20: Chi-square Comparison between the Models
Model comparison χ2 DF
Main – A 790.449 2
Main – B 618.24 59
Main – C 241.65 233
Main – D 545.97 224
13 See the discussion in Chapter Five – Section 5.9.1.2.
Chapter Eight – Multivariate Data Analysis
-249-
8.5 Hypotheses Testing14
The results of the SEM analysis indicated that the path loading between
consistency and trustworthiness was positive and significant (p < .05). This result
supported the view that if service organisations wish to create or enhance
trustworthiness in their services, their customers require consistency in the services
provided within the organisation. This is especially true when the service provider
matches the customers’ expectations, which is perceived to be the most important
factor in forming consistency. This is not entirely surprising bearing in mind the
In this study a total of nine hypotheses were generated and tested through
SEM (see Figure 8.5 above). The results indicated that:
H1: Consistency is positively correlated with trustworthiness.
H2: Competence is positively related with trustworthiness.
H3: Integrity is positively related with trustworthiness.
H4: Benevolence is positively correlated with trustworthiness.
H5: Value Alignment is positively correlated with trustworthiness.
H6: Communication is positively correlated with trustworthiness.
H7: Trustworthiness is positively related to attitudinal loyalty.
H8: Trustworthiness is positively related to behavioural loyalty.
H9: Behavioural loyalty is positively related to attitudinal loyalty (a prediction).
A detailed discussion of the hypotheses and their impact on acceptance of the general
model follows.
H1: Consistency is positively correlated with trustworthiness.
14 It’s important to note that this study tested the model in the hotel sector because it happened to be an interesting venue. Nevertheless, the results of the study will be generalised beyond the hotel context, even though the hypotheses will be discussed with regard to hotels.
Chapter Eight – Multivariate Data Analysis
-250-
nature of the services in the hotel sector, where customers often travel great distances
in order to use the service. Moreover, stratification of customer needs is also seen a
crucial factor; this can be achieved by fast service, efficient request handling and
quick check-in and check-out. The service provider’s reliability and stability in its
services is also a factor in enhancing consistency.
H2: Competence is positively related with trustworthiness.
This hypothesis examined the relationship between competence and
trustworthiness. The results of SEM demonstrated a strong positive relationship
between the two constructs.
Customers viewed competence from the point of view that service providers
should possess the right skills in order to be able to deliver the right service and
handle all requests efficiently. This might be important for this particular sample
because the data were collected in five star hotels, where customers’ expectation
might be higher than lesser star-ranked hotels.
H3: Integrity is positively related with trustworthiness.
The third hypothesis concerned the relationship between integrity and
trustworthiness, and the results from the analysis proved a positive correlation with p
(< .05. Customers would like the service provider to show fairness in transactions (no
extra charges or added taxes, sincerity and fair charges). Maintaining the promises
that have been made to customers is very important due to the travel involved and the
risk taken by customers in using the service provider. Moreover, showing respect for
the customers’ point of view is regarded as a significant element in creating an
impression of integrity.
H4: Benevolence is positively correlated with trustworthiness.
In this hypothesis, it was postulated that benevolence is positively related with
trustworthiness. The results from the SEM analysis supported this hypothesis,
revealing a positive relationship between the constructs (p < .05). According to the
findings, it can be argued that benevolence can be enhanced when the service provider
is open to customers’ needs, especially when they require additional services (for
Chapter Eight – Multivariate Data Analysis
-251-
example, medical care or special dietary requirements). At the same time the
providers should deliver the service in a caring manner with the customers’ best
interest in mind (not always calculated as being dependent on profit and cost).
H5: Value Alignment is positively correlated with trustworthiness.
The results from SEM showed that this hypothesis is accepted, with a positive
relationship between the constructs. Customers would prefer the service provider to
share the same values as they do, and be ethical when dealing with them.
Furthermore, customers favour the service provider where employees value their
customers and are interested in more than just generating extra profits.
H6: Communication is positively correlated with trustworthiness.
In hypothesis six, communication was linked positively with trustworthiness.
As with the rest of the hypotheses, the results from the analysis confirmed this
positive relationship between the constructs with significant p (< .05). Customers
prefer the service provider that communicates clearly, especially in the context of a
wide variety of customers who from various backgrounds and speak different
languages.
Informing customers immediately of any problems is seen as an important
factor, which also supports the general security level for customers (in hotels where
the region is politically unstable). From the customers’ perspectives, immediate
response to customers when contacted and always keeping two-ways communication
channels open are important factors for determining their trustworthiness.
H7: Trustworthiness is positively related to attitudinal loyalty.
The results from SEM analysis confirmed that the path from trustworthiness to
attitudinal loyalty was positive and significant (p < .05). The results suggest that
customers will pay a little extra in order to keep dealing with the service provider,
since they agreed that ‘I am likely to pay a little bit more for using the services’, ‘I
will continue to visit this service provider even if the service charges are increased
moderately’ and ‘I have strong loyalty to this service provider’. Nevertheless, the
service provider should not utilise this fact to exploit their customers (keeping the
Chapter Eight – Multivariate Data Analysis
-252-
importance of value alignment in mind). Based on the path diagrams, we can
conclude that attitudinal loyalty can be seen as a model outcome.
H8: Trustworthiness is positively related to behavioural loyalty.
Hypothesis eight examined the relationship between trustworthiness and
behavioural loyalty. The results from the SEM analysis confirmed the hypothesis with
significant p (< .05). Customers expressed a behavioural level of loyalty (Rundle-
Thiele and Bennett 2001), according to which they would recommend the service
provider to other people, try new services and say positive things about the service
provider. These can all be seen as a signs of behavioural loyalty (Rundle-Thiele and
Bennett 2001).
H9: Behavioural loyalty is positively related to attitudinal loyalty (a prediction).
The ninth hypothesis predicted that behavioural loyalty will ultimately lead to
attitudinal loyalty. However, the SEM results did not uphold this hypothesis; in fact,
the results showed a negative correlation between the constructs with a significant p
(< .05). The explanation can be drawn from the fact that different customer
classifications have different expectations from the service provider; some customers
are using the service provider simply because it fits with their budget, while other are
interacting with the service provider because of their attitudinal loyalty. The selected
sample contained a significant percentage (27%) of customers who had used the
service provider less than once a year (see Chapter Seven); this may suggest that this
percentage of customers might not have formed a higher level of loyalty, and their
results affected the total sample and the path coefficient between the two constructs.
Overall, however, trustworthiness as the model focal point has been approved
since it has been supported by the six proposed antecedent constructs. The model also
proved that the outcome of trustworthiness can be demonstrated by the measurement
of attitudinal and behavioural loyalty.
Chapter Eight – Multivariate Data Analysis
-253-
Summary
The SEM for this study was conducted via a two-step approach, with analysis
of the measurement model followed by the structural model. Confirmatory Factor
Analysis (CFA) was carried out to endorse the model’s fit; this was reported by
several goodness-of-fit-indices. The results of the CFA indicated that the
measurement model is valid after the definition of the underlying relationships
between the constructs and the observed items.
The structural model was implemented after validation in the first part of the
SEM process, and in turn it achieved an acceptable level. The path analysis clearly
defined the relationships among the constructs, and reported the path coefficients that
were used to test the study hypotheses.
The research proposed nine hypotheses; each hypothesis is linked to a
construct in the conceptualised model. The results from the SEM analysis approved
all the hypotheses excluding one predicted relationship.
The next chapter will discuss the general conclusions and provide a thorough
discussion on the key findings, as well as providing recommendations for academics
and practitioners.
Chapter Nine – Conclusions and Managerial Implications
-254-
CHAPTER NINE: CONCLUSIONS AND
MANAGERIAL IMPLICATIONS
9.1 Introduction
Few would deny that trustworthiness in relationships plays a vital role in
business, even though the definition of the term is, at times, masked in complexity
and contradictions within the different schools of thought in marketing research. Most
people have an instinctive awareness that trustworthiness is necessary for successful
relationships. The various strands of literature all suggest that trustworthiness is a
globally important concept across several disciplines, ranging from economic theory
to different psychological schools of thought. As discussed in previous chapters,
perceived trustworthiness has been and still is an important concept in marketing
because of the many benefits it brings to service organisations, such as lowering
operational costs and enhancing long-term relationships with customers.
As was noted in Chapter One (section 1.3), this thesis aims to explore
trustworthiness and its determinants in service organisations. Hotels were chosen as
the arena to explore these constructs, and Jordan was selected as a suitable country to
test the model presented in the thesis. The primary objectives of the thesis, set out in
Chapter One, are:
1. To develop a theoretically grounded framework and develop a scale to
measure the determinants of trustworthiness and its outcomes.
2. To investigate the factors that influences the development of trustworthiness
between a service provider and its customers.
3. To empirically examine the proposed model in the Jordanian hotel sector.
In order to achieve the set aim and objectives, this thesis asserts that
trustworthy behaviour can be demonstrated by several characteristics (i.e. integrity,
competence, value alignment, communication, benevolence and consistency). Effort
Chapter Nine – Conclusions and Managerial Implications
-255-
invested in such behaviour is well rewarded; an outcome which is illustrated in the
behaviour and/or the attitudinal loyalty shown by the organisation’s customers.
The research was conducted to theoretically develop and empirically test a
structural equation model to explore trustworthiness and its determinants within
service organisations. The model was tested in the Jordanian hotel sector. The
proposed hypotheses that attempted to identify the structural relationships between the
nine constructs in the model were examined via a series of analyses using SPSS and
AMOS statistical analysis programs.
The research aim and objectives were met by proposing and validating,
theoretically and empirically, an operational model. All the main hypothesised
relationships in the model were found positive and significant, in which they support
the proposed model1
The aim of this final chapter is to draw conclusions, discuss the relevance the
research findings and to point to the key managerial and theoretical implications of
the research findings. There will also be a discussion of the limitations of the thesis,
and to make some suggestions for future research. The chapter starts by providing a
summary of the different phases and of the results achieved. The main contribution of
. However, there are a number of drawbacks in the research,
such as, focusing only on one industry rather than examining the model in different
industries; these limitations are discussed in section 9.7.
The principle foundation for this study was based on the fact that
trustworthiness is an antecedent to trust, and the concept should, therefore, be applied
within service organisations rather than merely focusing on general organisational
trust. Additionally, trustworthiness is seen, in this thesis, as a characteristic which
contributes to the strength of interpersonal relationships, intra-organisational
relationships and inter-organisational relationships. The qualities of trustworthiness
can operate at different levels within an organisation. Trustworthiness is considered to
be the basis of a stable collective existence, i.e. it helps to avoid uncertainty in
relationships. Furthermore, trustworthiness can be viewed as a method for promoting
cooperation and reducing risk in a variety of relationship settings, for example, within
organisation departments and between employees and customers.
1 Full discussion in Section 9.3.
Chapter Nine – Conclusions and Managerial Implications
-256-
this research can be found in the hypothesised framework to measure trustworthiness
and apply it in service organisations.
In brief, it is evident from measuring trustworthiness that two of the perceived
characteristics of organisations, namely competence and value alignment, had the
greatest impact on trustworthiness. Therefore, both these constructs should be seen as
the key factors influencing trustworthiness. Some limitations of this research come
from the broad coverage of the model and its restriction to service organisations in
only one country. Future research should examine the proposed model of
trustworthiness across several different countries in order to investigate any cross-
cultural differences regarding perceived trustworthiness in service organisations.
9.2 General Findings and Discussion
Beginning with a discussion of the research question(s), this thesis then
provided an overview of the theoretical background and empirical studies that already
exist in the literature. The objective of the study was to develop a theoretical model
about trustworthiness, and to empirically test the factors (or constructs) that are likely
to affect the formation of trustworthiness between customers and a service provider.
The corollary constructs (the outcomes of the model) include behavioural and
attitudinal loyalty.
Closer examination of the relationships between the remaining observed
indicators for the construct showed that the impact of trustworthiness was measured
by four indicators that are related to the service provider’s perceived ability to be
trustworthy. For example, the provider should maintain a high level of
professionalism, respond caringly when contacted by customers, and be honest with
its customers. In particular, being honest with the customer is seen as having the
highest impact on trustworthiness.
In terms of the antecedents of trustworthiness, six constructs were proposed.
Those constructs were divided into two groups, high level and low level antecedents,
adapted from McAllister’s (1995) work on calculative and affective trust which refers
to emotional and calculated benefits. In this model the low level group was related to
calculated trust, while the high level constructs were related to affective trust.
Chapter Nine – Conclusions and Managerial Implications
-257-
The low level antecedents consisted of two constructs, consistency and
competence, and each of these was measured by four items. Consistency items
involved: showing fairness in transactions; consistency in service delivery; keeping
promises; and showing a high level of respect. Matching expectations and satisfying
needs were the two items which had the greatest impact on consistency. This could be
seen to be linked with the risk associated with a service transaction – in other words,
the customer wishes to receive something of an equivalent value to the amount he/she
has paid.
Competence was also measured with four items. These included: acquiring the
right skills to perform the service; being efficient; handling most of the customers’
requests; and being reliable. Similar to consistency, the lowest type of expectations
was the major influence on competence. Possession of adequate skills and being able
to handle customer requests were the most two important factors affecting
competence; this it is suggested that this was because customers want their basic
needs to be fulfilled first, before they are exceeded by the provision of extra services.
The high level antecedents were integrity, benevolence, communication and
value alignment; each of these was measured with four items. The result implied that
keeping promises is the most important element in building integrity for the service
provider. Customers believe that the service provider should not always aim to make
the maximum profit from a transaction; rather, the provider should also prioritise
other matters even though this may affect their profit (for example, arranging
celebrations for customers on special occasions).
In the case of communications, customers believed that the service provider
should always keep them informed of any problems with the service, as they prefer
being informed over not knowing what is happening with the service they purchased,
i.e. getting involved with the service provider, and hence, the service process. On
some occasions customers didn’t feel that the service provider was taking their
interests into account during the service encounter; therefore, customers viewed this
item as the most important when discussing benevolence.
Chapter Nine – Conclusions and Managerial Implications
-258-
Although most of these items could be seen as stating the obvious, the
majority of the interviewed customers clearly expressed that on many occasions their
service provider(s) had neglected these factors and did not consider these items of as
important when providing them with the service. This had led to a perception of the
service provider as untrustworthy.
Having confirmed the proposed relationships with the trustworthiness
construct, the overall measurement model was tested to see if the theoretical
measurement model fit the data. The CFA process used two split samples – a
calibration sample (n = 265) and a validation sample (n = 264). After analysis of the
calibration sample, the overall measurement model was not re-specified to describe a
better-fitting model for the data because the model achieved an acceptable fit,
therefore, there was no need to re-specify it. The theoretical measurement model was
then validated using the validation sample.
Ideally, the results from the validation sample should replicate the results of
the calibration sample. The theoretical model used for the calibration sample was the
same as that used for the validation sample; the final model was re-tested to see if it
described the validation data without any major re-specifications. During this test-
retest procedure, the indicators with comparatively high measurement errors and low
correlations in relation to the theoretical constructs were dropped. At the end, thirty-
six indicators remained to measure the nine constructs of the conceptual model. These
analyses were performed not only to confirm the uni-dimensionality of the constructs,
but also to clarify the observed indicators for the associated construct.
Chapter Nine – Conclusions and Managerial Implications
-259-
9.3 Research Hypotheses and Recommendations
As noted previously, a structural equation model for trustworthiness was
applied to test the research hypotheses and attempt to identify the structural
relationships between the constructs. Eight of the nine hypotheses proposed in this
study were supported, each generating a significant level in their respective
standardised coefficient scores. Detailed discussions of the research findings
associated with the individual research hypotheses are as follows. The hypotheses are
ranked according to their impact on trustworthiness, hence, their importance to the
research, using the path loading scores2
1- Value alignment
as a distinction factor.
Antecedents of Trustworthiness
The fifth proposed hypothesis (H5) states that value alignment is positively
correlated with trustworthiness. This hypothesis was supported by the structural
equation analysis. Furthermore, as the analysis indicated value alignment and
competence had the highest impact on trustworthiness, and, hence, is seen as one of
the most significant constructs in the model. However, because value alignment is one
of the higher order antecedents of trustworthiness, it is harder to create and control
this construct because the items related to value alignment require the service provider
to align their attitudes and behaviours (including any ethical issues) with their
customers. These items address such issues as the service provider sharing values
with their customers (for example, having a strong position towards environmental
issues and fair trade). Moreover, the service provider is expected to be ethical when
dealing with his/her customers, and is interested in more than just making a profit –
for instance, by supporting a local community and/or creating a community of
customers.
Two additional issues need to be raised when discussing value alignment.
Firstly, it is difficult to project these values to all customers through the normal
communication channels that service providers often employ (for example, online
newsletters and websites). Secondly, there are issues around the alignment of
2 See Chapter Eight – Section 8.4.1.
Chapter Nine – Conclusions and Managerial Implications
-260-
globalised and cross-cultural values to please all customers. These issues can be
solved by regular staff training and the creation of personalised relationships with
customers; here the concept of a part-time marketer is considered a key factor, by
which all the service provider’s employees (both front and back office) should be
trained to clearly deliver these values to customers, both on the organisation’s
premises and outside the organisation. The significance of value alignment is
demonstrated in the research findings of a number of academics, such as, Morgan and
Hunt (1994); Mayer and Davis (1999) and Siegtist et al. (2002).
2- Competence
The second hypothesis (H2) states that competence is positively related with
trustworthiness. The results from the structural equation analysis supported this
hypothesis, and showed that this construct (along with value alignment) has the
highest impact on trustworthiness. Therefore, in order for service providers to be
perceived as trustworthy, they should have adequate skills in order to be able to
perform the service. Customers expect these skills to signify a high level of
efficiency, because even though the service provider might manage to excel in the
provided service, a lack of knowledge about providing the right service will be
negatively perceived by customers. In other words, the service provider should
provide the right service first and on every occasion and then seek to distinguish
him/herself in any additional or extra services.
Staff training is also an important aspect of competence, so that the
organisation’s staff can be relied upon to handle most of customers’ requests. In cases
where customers might not feel comfortable requesting extra services (i.e. services
outside the provided one), the service provider should always seek to meet these
requests. These findings are supported in the literature (see for instance, Jarvenpaa et
al. 2000; Bhattacherjee 2002; Sirdeshmukh et al. 2002).
Chapter Nine – Conclusions and Managerial Implications
-261-
3- Communication
The sixth hypothesis in the proposed model (H6) states that communication is
positively correlated with trustworthiness. The results showed that customers prefer
the service provider to maintain open communication channels, through which they
expect to receive continuous and clear messages concerning the service(s) they have
purchased. In this matter, there is a strong emphasis on ‘clear’ communication since
customers may not speak the same language as the service provider; hence, the
service provider should seek alternative ways to ensure clarity of communication – for
example, providing messages in multiple languages. Another important element is
responding promptly when contacted by customers; if the service provider doesn’t
respond immediately, customers are less likely to perceive the organisation as
trustworthy. This is due to the customer not feeling that they are the focus of the
organisation’s attention. Several academics also agreed on the importance of
communication in business to consumer markets, see for instance, Dwyer et al.
(1987); Anderson and Narus (1990); Morgan and Hunt (1994) and Grönroos (2000).
4- Integrity
The third put forward hypothesis (H3) states that integrity is positively related
with trustworthiness. The findings supported this hypothesis; although the items used
to measure this concept were seen by one of the judges used in this study as ‘stating
the obvious’, customers clearly indicated that service provider(s) often do not perform
these basic service elements. Therefore, in order to achieve integrity in service
organisations, the service provider should keep his/her promises, show fairness when
dealing with customers and show no favouritism (or else customers will not perceive
the service provider as trustworthy). Even though service providers may employ
various loyalty programs and/or premier services, these should not interfere with or
influence regular customers (non-loyalty program users). This hypothesis is rooted in
the literature (see, for example, Mayer et al. 1995; Cummings and Bromiley 1996; De
Wulf et al. 2001).
Chapter Nine – Conclusions and Managerial Implications
-262-
5- Benevolence
The fourth hypothesis in the proposed model (H4) states that benevolence is
positively correlated with trustworthiness. This hypothesis indicates that benevolence
is significantly and positively correlated with trustworthiness. Benevolence in service
organisations can be expressed via seemingly insignificant characteristics in the
service provider; these characteristics should be initiated by the service provider, and
range from acting in a caring manner towards customers to having the customers’ best
interest in mind and being receptive to customers’ needs in all situations. This attitude
is exemplified in the basic service principle of ‘going the extra mile’, in which service
providers establish strong ties with customers by creating personalised relationships
with their customers and going beyond the image that the company only seeks profit
from their customers. These findings were supported in the existing literature (for
example, Mayer et al. 1995; Bhattacherjee 2002; Jarvenpaa et al. 2000).
6- Consistency
The first put forward research hypothesis (H1) states that consistency is
positively correlated with trustworthiness. The findings of the structural equation
analysis supported this hypothesis, finding out that there is indeed a positive
relationship between consistency and trustworthiness. Consequently, the results of the
analysis concluded that consistency in service organisations should be perceived as an
indication of stability in the overall provided service, i.e. the provided service should
maintain certain standards all the time and these standards should not be influenced or
negatively change with time. In addition, the service providers should always be in a
state that customers perceive as being reliable either by being professional or by
showing their competency skills. Moreover, the service provider should first strive to
satisfy customers’ needs by matching their expectations and only then attempt to
exceed customers’ expectations by providing additional value to the service, such as
enhanced service quality and improved customer service. These findings are aligned
with the previous literature (for instance, Hess 1995; Garbarino and Johnson 1999).
Chapter Nine – Conclusions and Managerial Implications
-263-
The Outcome of Trustworthiness
7- Attitudinal and Behavioural Loyalty
In the proposed model, trustworthiness was linked with attitudinal and
behavioural loyalty via the seventh hypothesis (H7) and the eighth hypothesis (H8).
These hypotheses state respectively:
H7: trustworthiness is positively related to attitudinal loyalty.
H8: trustworthiness is positively related to behavioural loyalty.
The model also suggests a prediction between behavioural loyalty and
attitudinal loyalty, by which behavioural loyalty is positively correlated with and
leads to attitudinal loyalty (hypothesis H9).
Customers believe that the service provider should have certain quality in
order to be perceived as trustworthy in addition to the mentioned antecedent above.
For example, the service provider should respond caringly when approached by
customers especially when dealing with problems or matters of urgency. Moreover,
the service provider should always be honest with customer and be professional and
dedicated to them.
Creating loyal customers is always a positive sign for organisations (see,
Reichheld and Sasser 1990). By being behaviourally loyal to the organisation,
customers are more likely to try new services and recommend the organisation to
others (i.e. an effective world of mouth). This level of loyalty should allow service
providers to further enhance their marketing activities through having customers
spreading positive word of mouth about the organisation and the experience they had.
Attitudinal loyalty is seen, in this study, as the highest form of loyalty in which
customers have strong sense of belonging to the organisation, wherein they are less
likely to switch to a different service provider even if the prices increase. This form of
loyalty is certainly the ultimate aim for any organisation as it will create a life time
customer.
Chapter Nine – Conclusions and Managerial Implications
-264-
The research findings supported H7 and H8, this is also shows in the work of
Dick and Basu (1994) and Rundle-Thiele and Bennett (2001). However, hypothesis
H9 was rejected because the causal relationship was not significant. Despite this
finding, this research will not disregard hypothesis H9 from the model since it may be
confirmed in a different setting or when it is applied and tested in a different industry.
Therefore, it is suggested that this hypothesis should receive more attention in future
studies to test its applicability.
Overall, the findings from the structural equation analysis supported all the
research hypotheses with the exception of H9. The analysis also clearly showed the
items that had the highest impact on their respective construct(s), and, hence, it
facilitated the model’s employment within a particular service organisation from a
managerial perspective. This possibility will be examined in more detail in the
following section.
Chapter Nine – Conclusions and Managerial Implications
-265-
9.4 Contribution of the Thesis
As noted earlier, the main contribution of this research is the establishment of
a conceptual model to measure trustworthiness, employing distinct determinants and
allowing the ability to predict the relationship outcome. By developing such a model,
it becomes possible to understand and apply the concept of trustworthiness in any
service organisation. On the theoretical side, trustworthiness was illustrated in detail
with respect to its antecedents; this was achieved by examining the various schools of
thought concerning trustworthiness in order to develop a marketing-specific model.
The empirical findings from the proposed model shows that the six proposed
determinants of trustworthiness do, in fact, have a strong positive impact on the
construct. In addition, trustworthiness has a positive impact on both behavioural and
attitudinal loyalty. Hence, it can be argued that the model is appropriate for modelling
trustworthiness within different settings; in other words, customer loyalty can be
achieved by applying the six determinants of trustworthiness.
In brief, the compelling contributions of the study can be summarised by the
following four key points:
• The research expands our knowledge of the concept of trustworthiness from a
marketing perspective. A clear distinction is drawn between trust and
trustworthiness. Moreover, trustworthiness is conceptualised along with a
rationale for using the concept of trustworthiness rather than trust.
• The study identifies the determinants of trustworthiness and examines two
types of customer loyalty. These are conceptualised as the outcome of
trustworthiness, therefore, a theoretical model is formed.
• The research goes on to apply a new scale development procedure to assess
the theoretical model; this was done by mainly using the C-OAR-SE scale
development procedure.
• Finally, the study tests the model empirically within a new service venue (the
hotel sector) and an unstudied region (the Middle East – Jordan).
The following sections will expand on these contributions and examine to
what extent they represent the achievement of the aims and objectives listed above.
Chapter Nine – Conclusions and Managerial Implications
-266-
9.4.1 Theoretical and Methodological Contributions
With respect to building theory, this study represents at least three distinct
contributions to the body of knowledge.
Firstly, because previous theories of relationship marketing, particularly
concerning trust and trustworthiness, are distributed among different domains (for
example, psychology and sociology), this study increases our understanding of the
concept of trustworthiness within the domain of marketing by clearly separating trust
from trustworthiness. The research highlights the fact that trustworthiness is an
antecedent of trust, in contrast to the various studies which suggest that trust alone is
the key factor in successful relationships.
Although the focus of this study is on the evaluation of trustworthiness, the
concept of customer loyalty was also included in order to establish a practical
outcome of trustworthiness. This focus has contributed to the body of knowledge by
increasing our understanding of the method by which two separate types of customer
loyalty (attitudinal and behavioural) are formed, and thus, by clarifying the ways that
trustworthiness is associated with customer loyalty in a single setting. Other
marketing literature, to date, has yet to conceptualise this relationship in a single
model.
Secondly, the evaluation of trustworthiness within services marketing has been
a neglected area in the existing marketing literature; it is argued that this neglect can
be attributed to the complex nature of trustworthiness. This study has increased our
knowledge about the formation of trustworthiness within service organisations by
creating, clarifying and applying the model within this sector. This provided various
insights on the hotel sector and how organisations can be improved by creating better
customer relationships through trustworthy behaviour initiated by the service
provider, rather than through a traditional reliance on customers to initiate such
behaviour.
Thirdly, this study can be seen as the first attempt to measure and empirically
test trustworthiness, using its antecedents and an outcome within the same model.
Further discussion on the contribution follows.
Chapter Nine – Conclusions and Managerial Implications
-267-
The three areas of contribution mentioned above, can be seen as the largest
areas in which this study has succeeded in making a contribution. However, there are
also several minor contributions arising from the study. For example, there is a
contribution to the theoretical discussion concerning the concepts of service quality
and service encounters; this study suggests that these concepts are not easily
separable, but work interchangeably to achieve better customer service (the scope of
this thesis did not cover these concepts due to the focus on the main construct of the
study). Relationship quality evaluation is present both in the content of the evaluation
(i.e. the economic factor) and in the adjustment processes (i.e. fairness).
The methodological contribution of the study relates to the construction of the
conceptual framework which can be used in both quantitative and qualitative studies.
In this study the conceptual model was built by a three step process, from the initial
preliminary framework, through a modified system to the final refined model. It is
argued that the refined framework represents the first attempt to conceptualise the key
constructs of the study – the determinants of trustworthiness and the different levels
of customer loyalty – in a manner which could in the future be applied to different
venues/sectors as well as to the hotel sector. The theoretical basis on which the
refined framework was partly built, together with the empirical evidence of its
validation, form reasonable justifications for its continued use in future studies.
Furthermore, this thesis can be seen as offering a methodological contribution
by developing and testing a new scale to empirically measure the theoretical model.
The examination of previous scales to measure the construct of trustworthiness as
well as customer loyalty led to the generation of new items that are more suitable for
use in service organisations. This provides future studies with guidance on how to
conceptualise trustworthiness within the domain of services marketing.
Moreover, the adoption of a new scale development procedure, the C-OAR-SE
procedure, is seen as a methodological contribution since it constitutes an empirical
test of this new approach, providing insights on its practicality and validity. From the
perspective of this thesis, applying the C-OAR-SE scale development procedure is
treated as an adequate method for developing the measurement scale for the study
because it grants the research the ability of relying on the well established theory (i.e.
content validity) over relying on data scores as had been advocated by Churchill
Chapter Nine – Conclusions and Managerial Implications
-268-
(1979). Still, several points have been taken into consideration when dealing with the
new procedure. One such example is Diamantopoulos’ (2005) argument regarding
scale adaptation which was used here in scale formation (for example, identifying the
judges and providing empirical evidence as well as content validity).
Chapter Nine – Conclusions and Managerial Implications
-269-
9.5 Managerial Implications of the Research Findings
In an increasingly competitive marketplace, an understanding of how
trustworthiness can be established and sustained may be a fundamental factor in
achieving a competitive edge for successful service organisations. Since the service
sector involves multi-faceted components, there is a need for a systematic analysis
and framework for the creation of trustworthiness. Such an analytical model may also
contribute to creating and integrating value added relationships in order to achieve
greater customer loyalty.
This study focused on an investigation of the structural relationships between
service organisation customers’ preferences about trustworthiness, what creates it (i.e.
its antecedents), and the possible outcomes of establishing a perception of
trustworthiness within service organisations. The most critical research finding from
this study was the strong relationship (represented by the highest path coefficient
score) between trustworthiness and value alignment and competence. Accordingly,
the following discussion of the managerial implications of this study is focused on
this finding, rather than on the influence of consistency and integrity (because of the
low impact of their scores on trustworthiness).
Importantly, these research findings will help service organisations and service
managers to understand how customers prefer to develop relationships with their
service provider, and under what circumstances they perceive the service provider to
be trustworthy. Therefore, the findings will help organisations and managers to plan
and implement successful and competitive strategies. These results will help service
organisations and marketers to collect information and plan appropriate strategies
based on the type of customer loyalty they prefer to develop.
For service managers, it would be advisable to address the issue of
trustworthiness as a starting point to assist in the process of achieving a high quality
of service. In contrast with the majority of the existing literature, this thesis suggests
that managers should focus on trustworthiness rather than trust, because trust is
initiated from the customer’s side, while trustworthiness is established from the
organisation’s side. Therefore, trustworthiness minimises the risk factor associated
with relying on customers to make the decision to trust the organisation. Instead, an
Chapter Nine – Conclusions and Managerial Implications
-270-
attempt to establish trustworthiness will create a trustworthy image for the
organisation, initially attracting customers to transact with the organisation, i.e. to
purchase their service, and subsequently helping to retain customers if the service
provider can demonstrate different aspects of trustworthiness (for example,
competence and benevolence) to customers during their service experience.
There should also be a distinction between high and low level antecedents of
trustworthiness, as suggested earlier in this chapter. Low level antecedents refer to
customers’ rational decision-making, while high level determinants are associated
with the emotional aspects of the relationship. Therefore, it could be suggested that
service providers should establish a strong customer relations foundation based on the
low level determinants (namely competence and consistency). This implies that
managers should educate their staff and provide full training programs regarding the
consistent delivery of good customer service over a period of time.
After establishing the cognitive or rational foundation for trustworthiness,
managers should start to focus on the higher level determinants, namely
communication, integrity, value alignment and benevolence. These constructs should
be addressed and established depending on the service provider’s individual
prioritisation; for instance, one manager might argue that communication is more
important than integrity for that particular organisation at a given moment.
Nevertheless, we can argue that these constructs should be treated on their path scores
(i.e. their impact on trustworthiness), which means that managers should consider
value alignment to be the most important factor for achieving trustworthiness,
followed by communication, then integrity and finally benevolence.
It could be concluded from the findings of this study that strategies for
building trustworthiness, supported by customer loyalty as a desirable outcome, can
be seen as a method for creating a competitive advantage for service organisations.
Such competitive strategies could be implemented based on the service provider’s
resources and overall business objectives. Organisations could either apply some of
the antecedents or all of them; for example, they could schedule regular staff training
on the specific factors that were thought to be important by their customers.
Chapter Nine – Conclusions and Managerial Implications
-271-
9.6 Directions for Future Research
Future research on trustworthiness should take place in different service
organisations to further confirm the findings of this research. Qualitative and
quantitative examination of the concept should both be undertaken, focusing in
particular on cross-industry analysis and cross-cultural differences between service
organisations.
Several questions emerge from the discussion of trustworthiness presented in
this thesis. For example, what are the specific factors that result in a loss of trust in
service organisations? Future research could identify and rank these factors according
to the impact they have on the environment of the service organisation, and how this
might impact on customer loyalty. A cross-cultural examination of the model is
important, in order to investigate the generalisation of the model across cultures;
because this model has been developed and tested in the Middle East, it could be
argued that a comparison of these results with a similar investigation in Europe or
Africa would be an important addition to the body of knowledge surrounding
trustworthiness. Furthermore, a cross-industry examination of the model is vital to if
the model is to be applicable across a range of service organisations, such as banks,
restaurants or tour operators.
Another question for further investigation could be the role of trustworthiness
in daily organisational life and how this is been measured. For example, in what ways
do individuals seek to confirm theories they hold about the trustworthiness of other
people prior to entering into exchanges with them? Furthermore, do these preliminary
tests of trustworthiness have the power to create suspicion in organisations
characterised by asymmetrical power relationships? If trust is seen to have a moral
dimension, what role could ethics play in the creation, maintenance and termination of
relationships from the management side? Finally, to what extent do brand names
impact the organisation’s perceived level of trustworthiness?
Chapter Nine – Conclusions and Managerial Implications
-272-
9.7 Study Limitations
As should be expected in all research, there were limitations to this study that
should be addressed to provide a basis for more sound research in the future. As has
already been discussed in the methodology chapter (see Chapter Five, section 5.12)
the major limitations affecting this study were: 1) research scope and boundaries of
the research; 2) the observed indicators and constructs’ 3) a lack of context
specificity; and 4) an absence of longitudinal characteristics.
This study investigated the structural relationships between trustworthiness
and its outcome and determinants. Data collection only took place in Amman, the
capital city of Jordan. This meant that the survey was geographically limited, and,
therefore, may have produced culturally specific results and conclusions in terms of
the magnitude and direction of the relationships among the constructs in the model.
Future research should address any differences between the Middle East (where this
study took place) and other continents.
Trustworthiness and customer loyalty in other countries may have different
drivers, and customers may reveal different perceptions, attitudes, and behaviours
concerning how they develop a perception of trustworthiness in their service provider.
Other cultures and geographical areas should be explored to see if similar findings
and results can be obtained. Also, future research could collect data from other service
industries (for example, banks) so that further comparison studies may be conducted.
In the current service sector, any service organisation may need to pay more
attention to advanced technologies and techniques so that high quality products and
services are delivered effectively and efficiently. In particular, the concept of self
service technologies is a rapidly growing area. Therefore, future studies should
address the antecedents of trustworthiness in relation to information technology and
self service technologies, and examine their impact on the different levels of customer
loyalty.
Another critical limitation of this study is related to the respondents.
Generally, in hotels, customers may have different perceptions of trust in the
organisation before and after they experienced the service. Therefore, it would be
Chapter Nine – Conclusions and Managerial Implications
-273-
extremely interesting to examine the development of their perceptions over time,
investigating their expression of trust before they experience the service offered, and
comparing this with their perceptions during and after the service encounter.
Accordingly, the conclusions that can be drawn from this study are somewhat
limited in terms of their longitudinal characteristics. Such data would it make possible
to analyse any potential time-lag for the hypothesised relationships and the structural
model. The data in the current study were collected over a three month period. A
longitudinal study might reflect ongoing transformations that could influence the
relationships between the constructs for future research, as well as further refining and
validating the model. However, cross-sectional studies are often based on capturing
the researched problem at a given moment, which only validates the research during a
certain period of time. Unlike cross-sectional studies, longitudinal studies track the
same research units (such as sales or consumption patterns) over a longer period of
time, and, therefore, longitudinal studies are useful for studying individual levels of
change over time, in contrast to cross-sectional datasets, which provide a snapshot of
a population at a single point in time (or at frequent intervals). Time, however, is itself
one of the most significant explanations of change. Therefore, longitudinal studies can
give answers to questions concerning change that cross-sectional studies cannot
provide.
This study did not include any commitment and satisfaction variables to see
how they relate to trustworthiness, its determinants and customer loyalty. Therefore,
future research should address this limitation to further enable the creation of
trustworthiness within the service sector. All the limitations mentioned above should
be considered as essential and critical suggestions for future research. Future studies
should take these limitations into account in order to produce more complete research
results.
In light of the fact that there is a limited number of empirical studies of
trustworthiness within the marketing domain, this study developed and empirically
tested a structural equation model of trustworthiness, taking into account its
determinants and outcome from the perspective of customers in the service sector.
Accordingly, as discussed in the research findings, it is hoped that this study has made
a valuable contribution to the understanding of the establishment of trustworthiness,
Chapter Nine – Conclusions and Managerial Implications
-274-
and has suggested some strategies for enhancing customer loyalty within service
organisations.
From the results of the comprehensive data analyses, this study concludes that
when a service organisation is interested in constructing trustworthiness by applying
the antecedents proposed in the model, it is vital to consider not only the outcome
(customer loyalty), but also the factors that influencing the antecedents. Nevertheless,
a more thorough understanding of trustworthiness and its components in the
marketing domain is necessary. Customers should always be seen as key players in
influencing the types of service and relationship they will enter into; moreover, it is
necessary to understand their preferences as well as their expectations for the types of
relationship they engage in, so that more effective relationship marketing can be
achieved.
Finally concluding, even though the results and the findings of this thesis are
somewhat confirmatory in nature, it is hoped that the information produced and the
implications of the study may be of help to service organisations, hotel managers, and
marketers, enabling them to create more competitive customer loyalty relationships
through the effective establishment and maintenance of their perceived
trustworthiness.
References
-275-
REFERENCES
Aaker, D., Kumar, V., Day, G., (2001), Marketing Research, John Wiley and Sons Inc: New York.
Aijo, S., (1996), The Theoretical and Philosophical Underpinnings of Relationship Marketing, European Journal of Marketing, Vol. 30(2), pp. 8-18.
Alreck., P., Settle, R., (1985), The Survey Research Handbook, Homewood.
Andaleeb, S., (1995), Dependence Relations and the Moderating Role of Trust: Implications for Behavioural Intintions in Marketing Channels, International Journal of Research in Marketing, 12(2), pp. 157-72.
Andaleeb, S., (1996), An Experimental Investigation of Satisfaction and Commitment in Marketing Channels: The Role of Trust and Dependence, Journal of Retailing, Vol. 72 (1), pp. 77-93.
Andaleeb, S., (1992), The Trust Concept: Research Issues for Channel of Distribution, in J.N. Sheth (ed.) Research in Marketing, Vol. 11, Jai Press Inc., pp.1-34.
Anderson E., Weitz B. (1989), Determinants of Continuity in Conventional Industrial Channel Dyads, Marketing Science, Vol.8 (fall), pp. 310-323.
Anderson J., Narus J., (1990), A Model of Distributor Firm and Manufacturer Firm Working Partnerships, Journal of Marketing, Vol. 54 (2), pp.42-58.
Anderson J., Gerbing D., (1988), Structural Equation Modeling in Practice: A Review and Recommended Two Step Approach, Psychological Bulletin, May, pp.411-423.
Anderson, L., Lodish L., Weitz B., (1987), Resource Allocation Behaviour in Conventional Channels, Journal of Marketing Research, Vol.24 (February), pp. 85-97.
Andrews, J., (2009), The Art of Persuasion, Nursing Standard, Vol. 23 (41), pp. 64.
Axelrod, R., (1984), The Evolution of Cooperation, New York: Basic Books.
Baier, A., (1986), Trust and Antitrust, Journal of Ethics, Vol. 96, pp. 231.
Baker, M., (2002), Research Methods, The Marketing Review, Vol. 3, pp. 167-193.
Baker, M., (2000), Selecting a Research Methodology, The Marketing Review, Vol. 1, pp. 373-397.
References
-276-
Barber, B., (1983), The Logic and Limits of Trust, New Brunswick. NJ: Rutgers University Press.
Barney, B., Hansen, H., (1994), Trustworthiness as a Source of Competitive Advantage, Strategic Management Journal, Vol.15, pp. 175–190.
Bartelt, A., (2002), Vertrauen in Zuliefernetzwerken, DUV, Wiesbaden, in Castaldo, S., (2007), Trust in Market Relationships, Edward Elgar, UK.
Baumgartner, J., (1991), Non-Marketing Professionals Need More than 4Ps, Marketing News, Vol. 22(July), pp. 28.
Bearden, O., Netemeyer, G., (1999), Handbook of Marketing Scales: Multi-item Measures for Marketing and Consumer Behavior Research, Sage: Thousands Oaks, CA.
Bentler, M., Bonnet, C., (1980), Significance Tests and Goodness of Fit in the Analysis of Covariance Structures, Psychological Bulletin, 88 (3), pp. 588-606.
Berry, L., (1983), Relationship Marketing, in Emerging Perspectives on Services Marketing, in: Berry, L., Shostack, L., Upah, D., (eds.), Chicago, American Marketing Association, pp. 25-28.
Berry, L., (1995), On Great Service: A Framework for Action, Free Press.
Berry, L., Conant, S., Parasuraman, A., (1991), A Framework For Conducting a Service Marketing Audit, Journal of the Academy of Marketing Science, Vol. 19,(Summer), pp. 255-68.
Berry, L., Parasuraman, A., (1993), Building a New Academic Field: the Case of Services Marketing, Journal of Retailing, Vol. 69(1), pp. 13-60.
Bews, F., Rossouw, J., (2002), A role For Business Ethics in Facilitating Trustworthiness, Journal of Business Ethics, Vol. 39(4), pp. 377-90.
Bhattacharya, R., Devinney, M., Pillutla, M., (1998), A Formal Model of Trust Based on Outcome, Academy of Management Review, 23 (3), pp. 459-472.
Bhattacherjee, A. (2002). Individual Trust in Online Firms: Scale Development and Initial Test, Journal of Management Information Systems, Vo. 19, pp. 211–241.
Bitner, M., (1990), Evaluating Service Encounters: The Effects of Physical Surroundings and Employee Responses, Journal of Marketing, Vol. 54(2), pp. 69-82.
Blois, K., (1998), Don’t all Firms Have Relationships?, Journal of Business and Industrial Marketing, Vol.13(3), pp. 256-270.
Blois, K., (1999), Trust in Business to Business Relationships: An Evaluation of its Status, Journal of Management Studies, Vol. 36(2), pp. 197-215.
References
-277-
Blomqvist, K., (1997), The Many Faces of Trust, Scandinavian Journal of Management, Vol. 13, pp. 271- 286.
Bollen, A., (1989), Structural Equations with Latent Variables, Toronto, ON, Canada: John Wiley and Sons, Inc.
Booms, H., Bitner, M., (1981), Marketing Strategies and Organisation Structures for Service Firms, Marketing of Services, Chicago American Marketing Association, pp. 47-51.
Boon, S., Holmes, J., (2001), The Dynamics of International Trust: Resolving Uncertinity in the Face of Risk, in Castaldo, S., (2007), Trust in Market Relationships, Edward Elgar, UK.
Borden, N., (1964), The Concept of the Marketing Mix, Journal of Advertising Research, June, pp. 2-7.
Borsboom, D., Mellenbergh, G., Heerden, J., (2004), The Concept of Validity, Psychological Review, Vol. 111(4), pp. 1061–1071.
Brenkert, G., (1998), Trust, Morality and International Business, in: Castaldo, S., (2007), Trust in Market Relationships, Edward Elgar, UK.
Brennan, R., (2001), An Essay on the History and Future of Reliability From the Perspective of Replications, Journal of Educational Measurement, Vol. 38, pp. 295-317.
Bromiley, P., Cummings, L., (1995), Transaction Costs in Organisations with Trust, Research on Negotiation in Organisations, Vol. 5, pp. 219-247.
Bryman, A., Bell, E., (2003), Business Research Methods, Oxford University Press.
Bryman, A., Cramer, D., (1999), Quantitative Data Analysis: A Guide for Social Scientists, Rout-ledge; London.
Butler, K., Cantrell, S., (1984), A Behavioural Decision Theory approach to Modeling Dyadic Trust in Superiors and Subordinates, Psychological Reports, Vol. 55, pp. 19-28.
Buttle, F., (1996), Relationship Marketing: Theory and Practice, Chapman, London.
Buttle, F., Burton, J., (2002), Does Service Failure Influence Customer Loyality?, Journal of Consumer Behaviour, Vol. 1(3), pp. 17-27.
Büttner, B., Göritz, S., (2008), Perceived Trustworthiness of Online Shops, Journal of Consumer Behaviour, Vol. 7 (1), pp. 35-50.
Byrne, B., (1995), One Application of Structural Equation Modelling from Two Perspectives: Exploring the EAS and LISRELM Strategies. In R. H. Hoyle (ed.), Structural Equation Modelling: Concepts, issues, and Application (pp. 138-157). Thousand Oaks, CA: Sage.
Byrne, B., (1998), Structural Equation Modeling with LISREL, PRELIS, and SIMPLIS: Basic Concepts, Applications, and Programming, Lawrence Erlbaum Associates London.
References
-278-
Byrne, B., (2001), Structural Equation Modeling, with AMOS: Basic Concepts, Applications and programming, Lawrence Erlbaum Associates, London.
Caldwell, C., Clapham. E., (2003), Organisational Trustworthiness: An International Perspective, Journal of Business Ethics. Vol. 47, pp. 349-364.
Caldwell, C., Hayes, L. A., Karri, R., Bernal, P., (2008), Ethical Stewardship Implications for Leadership and Trust, Journal of Business Ethics, Vol. 78, pp. 153-164.
Carver, S., Scheier, F., (1998), On the Self-Regulation of Behaviour, New York: Cambridge University Press.
Castaldo, S., (2003), Trust Variety – Conceptual Nature, Dimensions and Typologies, IMP 2003 Comference, Lugano, Switzerland.
Castaldo, S., (2007), Trust in Market Relationships, Edward Elgar, UK.
Castelfranchi, C., Falcone R., (2001), Social Trust: A Cognitive Approach, in Trust and Deception in Virtual Societies, Kluwer Academic Publishers, pp. 55-90.
Cavana, R., Delahaye, B., Sekaran, U., (2001), Applied Business Research: Qualitative and Quantitative Methods, John Wiley and Sons Australia Ltd.
Chaudhuri, A., Holbrook, M., (2001), The Chain of Effects from Brand Trust and Brand Affect to Brand Performance: The Role of Brand Loyalty, Journal of Marketing, 65(2), pp. 81-93.
Christopher, M., (1996), From Brand Values to Customer Values, Journal of Marketing Practice: Applied Marketing Science, Vol. 2(1), pp. 55-65.
Christopher, M., Payne, A., Ballantyne, D., (1991), Relationship Marketing, Butterworth-Heinemann: Oxford.
Churchill, A., (1979), A Paradigm for Developing Better Measures of Marketing Construct, Journal of Marketing Research, Vol.16 (February), pp.64-73.
Clark, C., Payne, R., (1997), The Nature and Structure of Worker’s Trust in Management, Journal of Organisational Behaviour, Vol. 18, pp. 205-224.
Clark, K., Sefton, M., (2001), The Sequential Prisoner’s Dilemma: Evidence on Reciprocation, EconomicJournal, Vol. 111, pp. 51-68.
Cohen, T., Duggar, K., Gray, G., Kreindel, S., Abdelrahman, H., HabteMariam, T., Oryang, D., Tameru, B., (2001), Evaluation of the Potential for Bovine Spongiform Encephalopathy in the United States. Harvard Center for Risk Analysis/Harvard School of Public Health and Center for Computational Epidemiology, College of Veterinary Medicine, Tuskegee University. Available at http://www.aphis.usda.gov/lpa/issues/bse/madcow.pdf.
Coleman, S., (1990), Foundations of Social Theory, Cambridge, Harvard University Press.
References
-279-
Collis, J., Hussey, R., (2003), Business Research: A Practical Guide, 2nd Edition, Palgrave Macmillan.
Cook, C., Heath, F., Thompson, R., (2000), A Meta Analysis of Response Rates in Web or Internet Based Surveys, Educational and Psychological Measurement, Vol. 60, pp. 821-83.
Corritore, L., Kracher, B., Wiedenbeck, S., (2003), On-line Trust: Concepts, Evolving Themes, A Model, International Journal of Human–Computer Studies, Vol. 58(6), pp. 737–758.
Cramer, D., (2003), Advanced Quantitative Data Analysis, Open University Press.
Creswell, W., (1998), Qualitative Inquiry and Research Design; Choosing Among Five Traditions, London, Sage.
Creswell, W., (1994), Research Design: Qualitative and Quantitative Approaches, Thousand Oaks: Sage.
Creswell, W., (2003), Research Design: Qualitative, Quantitative and Mixed Methods Approaches, Thousand Oaks, CA: Sage.
Cronbach, J., (1969), Essentials of Psychological Testing, Harper and Row: New York.
Cronbach, J., (2004), My Current Thoughts on Coefficient Alpha and Successor Procedures, Educational and Psychological Measurement, Vol. 64, pp. 391-418.
Crosby, A., Evans, R., Cowles, D., (1990), Relationship Quality in Services Selling: An Interpersonal Influence Perspective, Journal of Marketing, 54 (July), pp.68-81.
Crosby, A., Stephens, N., (1987), The Effects of Relationship Marketing on Satisfaction, Retention and Prices in the Life Assurance Industry, Journal of Marketing Research, Vol. 24, pp. 404-411.
Crossan, F., (2002), Research Philosophy Towards an Understanding, Nurser’s Research, Vol.1 (1), pp. 46-55.
Culliton, W., (1948), The Management of Marketing Costs, Graduate School of Business Administration, Boston, Mass: Harvard University.
Cummings, L., Bromiley, P., (1996), The organisational trust Inventory (OTI), Development and validation in Castaldo, S., (2007), Trust in Market Relationships, Edward Elgar, UK.
Curral, S., Judge, T., (1995), Measuring Trust between Organisational Boundary Role Persons, Organisational Behaviour and Human Decision Processes, Vol. 65, pp. 601-620.
Czaja, R., Blair, J., (1996), Designing Surveys: A Guide to Decision and Procedure, Thousand Oaks, Pine and Forge Press.
References
-280-
Dasgupta, P., (1988), Trust as a Commodity, In Diego Gambetta, (ed), Trust, Oxford: Basil.
De Wulf, K., Gaby Odekerken-Schröder, I., (2001), Investments in Consumer Relationships: A Cross- Country and Cross-Industry Exploration, Journal of Marketing, Vol. 65 (October), pp. 33–50.
Denzin, K., (1970), The Research Act: A Theoretical Introduction to Sociological Methods, Aldine: Chicago.
Denzin, K., Lincoln, S., (1998), The Landscape of Qualitative Research, London, Sage.
Denzin, K., Lincoln, S., (2000), Handbook of Qualitative Research, 2nd edition, Sage Publications: London.
Deutsch, M., (1960), The Effect of Motivational Orientation Upon Trust and Suspicion, Human Relations, Vol. 13, pp. 123-140.
Deutsch, M., (1973), The Resolution of Conflict: Constructive and Destructive Processes, Yale University Press, New Haven, Connecticut.
DeVellis, F., (1991), Scale Development: Theory and Applications, Applied Social Research Methods Series, Vol. 26, Newbury Park: Sage.
DeVellis, F., (2003), Scale Development, 2nd edition, Thousand Oaks, CA: Sage Publications.
Diamantopoulos, A., (2005), The C-OAR-SE Procedure for Scale Development in Marketing: A Comment, International Journal of Research in Marketing, Vol. 22, pp.1-9.
Diamantopoulos, A., Winklhofer, M., (2001), Index Construction with Formative Indicators: an Alternative to Scale Development, Journal of Marketing Research, Vol. (38), pp. 269-277.
Dick, A, Basu, K., (1994), Customer Loyalty Towards an Integrated Framework, Journal of The Academy of Marketing Science, Vol. 22(2), pp. 99-113.
Doney, M., Cannon, P., Mullen, R., (1998), Understanding the Influence of National Culture on the Development of Trust, Academy of Management Review, Vol. 23, pp. 601-620.
Doney, M., Cannon, P., (1997), An Examination of the Nature of Trust in Buyer-Seller Relationships, Journal of Marketing, Vol. 61(April), pp.35-51.
Dunn, H., (1988), Trust and Political Agency, In D. Gambetta, Trust: Making and Breaking Cooperative Relations, pp. 73-93. New York: Blackwell.
Dwyer, F., Oh, S., (1987), Developing Buyer-Seller Relationships, Journal of Marketing, Vol. 51(2), pp. 11-27.
References
-281-
Dwyer, F., Schurr, H., Oh, S., (1997), Output Sector Munificence Effects on the Internal Political Economy of Marketing Channels, Journal of Marketing Research, Vol. 24, pp. 347–358.
Easterby-Smith, M., Thorpe, R., Lowe, A., (1991), Management Research: An Introduction, Sage Publications: London.
Easterby-Smith, M., Thorpe, R., Lowe, A., (1997), Management Research, 2nd edition, London: Sage Publications Ltd.
Edwards, D., (1998), Types of Case Study Work: A Conceptual Framework for Case Based Research, Journal of Humanistic Psychology, Vol. 38, pp. 36-71.
Edwards, J., Bagozzi, R., (2000), On the Nature and Direction of Relationships between Constructs and Measures, Psychological Methods, Vol. 5(2), pp. 155-74.
Eisenberger, R., Peter, F., Davis-LaMastro, V., (1990), Perceived Organisational Support and Employee Diligence, Commitment, and Innovation, Journal of Applied Psychology, Vol. 75, pp. 51-59.
Embretson, S., (1983), Construct Validity: Construct Representation versus Nomothetic Span, Psychological Bulletin, Vol. 93, pp. 179-197.
Ennew, C., Sekhon, H, (2003), Trust Marketing Relationships: A Conceptual Framework, Academy of Marketing Conference Procedure: UK.
Ennew, C., Sekhon, H, (2005), Trustworthiness in Financial Services: A Conceptual Framework and Scale Development, 34th EMAC, Bocconi University, Milan.
Eyles, J., (1989), Qualitative Method, in: Johnston, R., Gregory, D., Smith, D., (2000), The Dictionary of Human Geography, Oxford: Blackwell.
Farrell, D., Rusbult, C., (1981), Exchange Variables as Predictors of Job Satisfaction, Job Commitment, and Turnover: The Impact of Rewards, Costs, Alternatives, and Investment, Organisational Behaviour and Human Performance, Vol. 28, pp. 78-95.
Finn, A., Kayande, U., (2005), How Fine is C-OAR-SE? A Generalisability Theory Perspective on Rossiter’s Procedure, International Journal of Research in Marketing, Vol. 22, pp.11-21.
Fisk, R., Brown, S., Bitner, M., (1994), Tracking the Evolution of Services Marketing Literature, Journal of Retailing, Vol. 69(1), pp. 61-103.
Fiske, A., (1990), Relativity Within Moose Culture; Four Incommensurable Models for Social Relationships, Ethos, Vol. 18 pp.180-204.
Floyd, J., Fowler, Jr., (1993), Survey Research Methods, Sage Publications: London.
References
-282-
Flynn, L., Pearcy, D., (2001), Four Subtle Sins in Scale Development: Some Suggestions for Strengthening the Current Paradigm, International Journal of Market Research, Vol. 43(4), pp. 409-423.
Foddy, W., (1993), Constructing Questions for Interviews and Questionnaires: Theory and Practice in Social Research, Cambridge Press: UK.
Ford, D., Gadde, L., Håkansson, H., Lundgren, A., (1998), Managing Business Relationships, John Wiley and Sons: England.
Fornell, C., Larcker, D., (1981), Evaluating Structural Equation Models with Unobservable Variables and Measurement Rrror, Journal of Marketing Research, Vol. 18, pp. 39-50.
Fornell, C., Yi, Y., (1992), Assumptions of the Two-step Approach to Latent Variable Modelling, Socio- Logical Methods and Research, Vol. 20, pp. 291–320.
Foxall, G., (1998), Radical Behaviorist Interpretation: Generating and Evaluating an Account of Consumer Behavior, The Behavior Analyst, Vol. 21 pp.321-54.
Friedland, N., (1990), Attribution of Control as a Determinant of Cooperation in Exchange Interactions, Journal of Applied Social Psychology,Vol. 20, pp. 303-320.
Fukuyama, F., (1995), Trust: The Social Virtues and the Creation of Prosperity, New York: Free Press.
Gambetta, D., (1988), Trust; Making and Breaking Co-operative Relations, Basil Blackwell Ltd: Oxford.
Gambetta, D., (1989), Possiamo Fidarci Della Fiducia?, in Castaldo, S., (2007), Trust in Market Relationships, Edward Elgar, UK.
Ganesan, S., (1994), Determinants of Long-term Orientation in Buyer-Seller Relationships, Journal of Marketing, Vol. 58(2), pp.1 -19.
Garbarino, E., Johnson, M., (1991), The Different Roles of Satisfaction, Trust, and Commitment in Customer Relationships, Journal of Marketing, Vol. 63(2), pp. 70-87.
Garfinkel, H., (1967), Studies in Ethnomethodology, Englewood Cliffs, Prentice Hall.
Gay, R., Airasian, P., (1999), Educational Research: Competencies for Analysis and Application, 6th edition, Upper Saddle River: Merrill.
Gefen, D., Karahanna, E., (2003), Trust and TAM in Online Shopping: An Integrated Model, MIS Quarterly 27(1), pp. 51-90.
Gefen, D., (2000), E-commerce: The Role of Familiarity and Trust, Omega, 28 (5), pp. 725-37.
References
-283-
Gerbing, D., Anderson, J., (1985), The Effects of Sampling Error and Model Characteristics on Parameter Estimation for Maximum Likelihood Confirmatory Factor Analysis, Multivariate Behavioral Research, Vol. 20, pp. 255–271.
Ghoshal, S., Bartlett, C., (1994), Linking organisational Context and Managerial Action: the Dimension of Quality Management, Strategic Management Journal, Vol. 15 pp.91-112.
Gibb, R., (1964), Climate For Trust Formation in Castaldo, S., (2007), Trust in Market Relationships, Edward Elgar, UK.
Giffin, K., (1967), The Contribution of Studies of Source Credibility to a Theory of Interpersonal Trust in the Communication Department, Psychological Bulletin, Vol. 68, pp. 104-120.
Glaeser, E., Laibson, D., Scheinkman, J., Soutter, C., (1999), What is Social Capital? The Determinants of Trust and Trustworthiness, NBER Working Paper No. 7216.
Goldsmith, E., (1999), The Personalised Marketplace: Beyond the 4Ps, Marketing Intelligence and Planning, Vol. 17(4), pp. 178-185.
Govier, T., (1994), Is it a Jungle Out There? Trust, Distrust and the Construction of Social Reality, Dialogue, Vol. 33(2), pp.237-52.
Granovetter, M., (1985), Economic Action and Social Structure: The Problem of Embeddedness, American Journal of Sociology, Vol. 91, pp. 481-510.
Grazioli, S., Wang, A., (2001), Looking Without Seeing: Understanding Unsophisticated Consumers’ Success And Failure TO Detect Internet Deception, Twenty-Second International Conference on Information Systems.
Grix, J., (2004), The Foundations of Research, Palgrave Study Guides.
Grönroos, C., (2000), Creating a Relationship Dialogue: Communication, Interaction and Value, the Marketing Review, Vol. 1, pp. 5-14.
Grönroos, C., (1978), The Nature of Service Marketing, Working Papers, No. 11, Swedish School of Economics and Business Administration, Helsinki, Finland.
Grönroos, C., (1983), Strategic Management and Marketing in the Service Sector, Studentlitteratur, Sweden, and Chartwell-Bratt Ltd, UK.
Grönroos, C., (1985), Internal marketing - theory and practice, In Bloch, T.M., Block, Upah, G.D. and Zeithaml, V. (Eds), Services Marketing in a Changing Environment, American Marketing Association, pp.41-47.
Grönroos, C., (1993), Toward a Third Phase in Service Quality Research: Challenges and Future Directions, Swartz, T.A., Bowen, D.E., Brown, S.W., Advances in Services Marketing and Management, Vol. 2.
References
-284-
Grönroos, C., (1998) Marketing Services: the Case of a Missing Product, Journal of Business and Industrial Marketing, Vol. 3(4/5), pp. 322-338.
Grönroos, C., (2006), Adopting a Service Logic for Marketing, Marketing Theory, Vol. 6(3), pp. 317-333.
Grönroos, C., Ravald, A., (1996), the Value Concept and Relationship Marketing, European Journal of Marketing, Vol. 30(2), pp. 19-30.
Groves, R., Couper, M., Presser, S., Singer, E., Tourangeau R., Acosta, G., Nelson, L., (2006), Experiments in Producing Non-Response Bias, Public Opinion Quarterly, Vol. 70 (5), pp. 720-736.
Guba, E., Lincoln, Y., (1994), Competing Paradigms in Qualitative Research, In Denzin, N. K. and Lincoln, Y. S. (Eds.), Handbook of Qualitative Research, Thousand Oaks: Sage Publications, pp. 105-117.
Guest, L., (1944), A Study of Brand Loyalty, Journal of Applied Psychology, Vol. 28 pp.16-27.
Gummesson, E., (1978), Toward a Theory of Professional Service Marketing, Industrial Marketing Management, Vol. 7 (2), pp. 89–95.
Gummesson, E., (1991), Qualitative Methods in Management Research, Revised ed., Newbury Park: Sage: Chapter 1.
Gummesson, E., (1994), Marketing Relationship Marketing Operational, International Journal of Service Industry Management, Vol. 5(5), pp.5-20.
Gummesson, E., (1996), Relationship Marketing and Imaginary Organisations: a Synthesis, European Journal of Marketing, Vol. 30 No. 2, pp. 31-44.
Gummesson, E., (1997), Relationship Marketing as Paradigm Shift: Some conclusions from the 30R Approach, Management Decision, Vol. 35(4), pp. 267-272.
Hair, J., Anderson, R., Tatham, R., Black, B., (1998), Multivariate Data Analysis,sixth edition, Prentice Hall.
Hair, J., Babib, B., Anderson, R., Tatham, R., Black, B., (2006), Multivariate Data Analysis, eighth edition. Prentice Hall.
Håkansson, H., (1982), International Marketing and Purchasing of Industrial Goods: an Interaction Approach, John Wiley and Sons, Inc., Chichester, UK.
Håkansson, H., Johanson, J., (1992), A Model of Industrial Networks, in Axelsson, B., Easton, G. (Eds) Industrial Networks – A New Reality, Routledge, London.
Handfield, R., Bechtel, C., (2004), Trust, Power, Dependence, and Economics: Can SCM Research Borrow Paradigms? International Journal of Integrated Supply Management, Vol. 1(1), pp. 3-32.
References
-285-
Hardin, R., (2002), Trust and Trustworthiness, Russell Sage Foundation, New York.
Harker, M., (1999), Relationship Marketing Defined: And Examination of Current Relationship Marketing Definitions, Journal of Marketing Intelligence and Planning, Vol. 17(1), pp. 13-20.
Hart, M., Capps, R., Cangemi, P., Caillouet, M., (1986), Exploring Organisational Trust and its Multiple Dimensions: A Case Study of General Motors, Organisation Development Journal, Vol. 4(2), pp. 31-39.
Harvey, M., Lush, R., Cavarkapa, B., (1996), A Marketing Mix For the 21st Century, Journal Marketing Theory and Practice, pp. 1-15.
Hattie, J., (1985), Methodology Review: Assessing Uni-dimensionality of Tests and Items, Applied PsychologicalMeasurement, Vol. 9 (June), pp. 139-64.
Hayduk, L., (1996), LISREL: Issues, Debates, and Strategies, Baltimore: Johns Hopkins University Press.
Hess, J., (1995), in Stern, B.B. (Eds),Construction and Assessment of a Scale to Measure Consumer Trust, American Marketing Association, Chicago, IL, Vol. 6 pp.20-6.
Hesse-Biber, S., Leavy, P., (2004), Approaches to Qualitative Research: A Reader on Theory and Practice, Oxford University Press.
Hinkin, R., (1995), A Review of Scale Development Practices in the Study of Organisations, Journal of Management, Vol. 21(5), pp. 967-988.
Hodgson, M., (1988), Economics and Institutions, Cambridge: Polity Press.
Hollis, M., (1996), the Philosophy of Social Science: an Introduction, Cambridge: Cambridge University Press.
Horn, J., (1965), A Rationale and a Test for the Number of Factors in Factor Analysis, Psychometrika, Vol. 30, pp. 179–185.
Hosmer, L., (1995), Trust: the Connecting Link between Organisational Theory and Philosophical Ethics. Academy of Management Review, Vol. 20(2), pp. 379-403.
House, C., (1985), Questionnaire Design With Computer Assisted Telephone Interviewing, Journal of Official Statistics, Vol. 1 (2), pp. 209 – 219.
Hoyle, R., (1995), Structural Equation Modeling: Concepts, Issues, and Applications, Sage Publications.
Huemer, L., (2000), Beyond Predictability: Employing Trust in Business Relationships, in Castaldo, S., (2007), Trust in Market Relationships, Edward Elgar, UK.
Huemer, L., (1998), Trust in Business Relationships: Economic Logic or Social Interaction? Sweden, Borea Bokforlag.
References
-286-
Huysamen, K., (1997), Parallels Between Qualitative Research and Sequentially Performed Quantitative Research, South African Journal of Psychology, Vol. 27, pp. 1-8.
Hycner, H., (1999), Some Guidelines for the Phenomenological Analysis of Interview Data. In. Bryman, A., Burgess, R., Qualitative research, Vol. 3, pp. 143-164. London: Sage.
Iriondo, M., Albert, J., Escudero, A., (2003), Structural Equation Modeling: An Alternative for Assessing Causal Relationships in Threatened Plant Populations, Biol Conserv, Vol. 113, pp. 367-377.
Jarvenpaa, L., Tractinsky, N., Vitale, M., (2000), Consumer Trust in an Internet Store Information Technology and Management, Vol.1 (1-2), pp. 45-71.
Johanson, J., Mattsson, G., (1985), Marketing Investments and Market Investments in Industrial Networks, International Journal of Research in Marketing, No. 4, pp. 185-95.
Johns, N., (1999), What is Thing Called Service, European Journal of Marketing, Vol. 33(9/10), pp. 958-973.
Johnson, A., (1986), Adding More Ps to the Pod or 12 Essential Elements of Marketing, Marketing News, Vol.11, pp. 2.
Johnson-George, C., Swap, W., (1982), Measurement of Specific Interpersonal Trust: Construction and Validation of a Scale to Assess Trust in a Specific Other, Journal of Personality and Social Psychology, Vol. 43, pp. 1306-1317.
Jöreskog, G., (1973), A General Method for Estimating a Linear Structural Equation System, http://impgroup.org/paper_view.php?viewPaper=4317.
Jöreskog, K., (1993), Testing Structural Equation Models, In Bollen, K., and Long, S (Eds.), Testing Structural Equation Models (294-316). Newbury park, CA: Sage.
Jöreskog, K., Sörbom, D., (1996), PRELIS 2: User’s Reference Guide, Scientific Software International, Chicago.
Judd, C., (1987), Differentiate with the 5th P: People, Industrial Marketing Management, Vol. 16, pp. 241-247.
Jyh-Shen, C. (2004). The Antecedents of Consumers' Loyalty Toward Internet Service Providers, Information and Management, Vol. 41(6), pp. 685-690.
Kent, R., (1986), Faith in the Four Ps: An Alternative, Journal of Marketing Management, Vol. 2, pp. 145-155.
Kline, R., (1998), Principles and Practice of Structural Equation Modeling, NY: Guilford Press.
Kline, R., (1998), Software Programs for Structural Equation Modelling: AMOS, EQS, and LISREL, Journal of Psychoeducational Assessment, Vol. 16, pp. 343-364.
References
-287-
Kline, R., (2005), Principles and Practice of Structural Equation Modeling, NY: Guilford Press.
Koller, M., (1988), Risk as a Determinant of Trust, Basic and Applied Social Psychology, Vol. 9(4), pp. 268.
Korczynski, M., (1994), Low Trust and Opportunistic Action Evidence on Inter-Firm Relations from the British Engineering Construction Industry, Journal of Industry Studies, Vol. 1(2).
Korczynski, M., (2000), The Political Economy of Trust, Journal Management Studies, Vol. 37(1), pp. 21.
Kotler, P., Levy, S., (1969), Broadening the Concept of Marketing, Journal of Marketing, Vol. 33(1), pp. 10-15.
Kotler, P., (1972), a Generic Concept of Marketing, Journal of Marketing, Vol. 36(3), pp. 46-54.
Kotler, P., (1986), Mega-Marketing, Harvard Business Review, Vol. 64, pp.117-24.
Kotler, P., (2003), Marketing Management, 11th Edition, Prentice Hall, New Jersey.
Kumar, N., Scheer, L., Steenkamp, J.-B., (1995), The Effects of Perceived Interdependence on Dealer Attitudes, Journal of Marketing Research, Vol. 32(3), pp. 348-356.
Lane, C., Bachmann, R., (1998), Trust within and between Organisations: Conceptual Issues and Empirical Applications. Oxford University Press, Oxford.
Langeard, E., Eiglier, P., (1987), Servuction: Les Marketing des Services, John Wiley, Paris.
Larzelere, R., Huston, T., (1980), The Dyadic Trust Scale: Toward Understanding Interpersonal Trust in Close Relationships, Journal of Marriage and the Family, Vol. 42, pp. 595-604.
Lee Goi, C., (2009), A Review of Marketing Mix: 4Ps or More?, International Journal of Marketing Studies, Vol 1(1). Pp. 2-15.
Levitt, T., (1960), Narketing Myopia, Harvard Business Review.
Levitt, T., (1983), After the Sale is Over, Harvard Business Review, September – October.
Lewicki, R. , Bunker B., (1996), Developing and Maintaining Trust in Work Relationships, in R. Kramer and T. Tyler (eds.) Trust in Organizations: Frontiers of Theory and Research, Sage Publications, Thousand Oaks, California, pp. 114-139.
Lewicki, R., Bunker, B., (1995), Trust in Relationships: A Model of Trust Development and Decline, in B. Bunker and J. Rubin, Conflict and Justice, San Francisco: Jossey-Bass.
References
-288-
Lewicki, R., J., McAllister D., Bies, R., (1998), Trust and Distrust: New Relationships and Realities, Academy of Management Review, 23 (3), pp. 438-458.
Lewis, D., Weigert, A., (1985) Trust as a Social Reality, Social Forces, Vol. 63, pp. 967-985.
Liljander, V., Strandvik, T., (1996), Emotions in Service Satisfaction, International Journal of Service Industry Management, Vol. 8(2), pp. 148-169.
Lorenz, E., (1988), Neither Friends nor Strangers: Informal Networks of Subcontracting in French Industry. In Gambetta, D., (ed.), Trust: Making and Breaking Cooperate Relations, Oxford: Blackwell.
Luhmann, N., (1988), Familiarity, Confidence, Trust: Problems and Alternatives. In D. G., Gambetta (Ed.), Trust, pp. 94-107, New York: Basil Blackwell.
Macintosh, G., (2002), Perceived Risk and Outcome Differences in Multi-Level Service Relationships, Journal of Services Marketing, Vol. 16(2), pp.143-157.
Macionis, J., Plummer, K., (1997), Sociology, A Global Introduction, Prentice Hall Inc: New Jersey.
Mackie, D.M., and Hamilton, D.L., (1993), Affect, Cognition and Stereotyping: Interactive Processes in Group Perception, Academic Press.
Macy, M., Skvoretz, J., (1998), The Evolution of Trust and Cooperation Between Strangers: A Computational Model, American Sociological Review, Vol. 63, pp. 638-660.
MaGrath, A., (1986), When Marketing Services, 4Ps Are Not Enough, Business Horizons, Vol. 29(3), pp. 45-50.
Malaeb, A., Summers, K., Pugesek H., (2000), Using Structural Equation Modeling to Investigate Relationships Among Ecological Variables, Environ Ecol Stat, Vol. 7, pp. 93-111.
Malaga, A., (2001), Web Based Reputation Management Systems: Problems and Suggested Solutions, Electronic Commerce Research, Vol. 1, pp. 403-17.
Malecki, E., Tootle, D., (1996), The Role of Networks in Small Firm Competitiveness, International Journal of Technology Management, Vol. 11(1/2), pp. 43-58.
Maruyama, M., (1998), Basics of Structural Equation Modeling, California: Sage.
Mathe, H., Shapiro, R., (1993), Integrating Service Strategy in the Manufacturing Company, Chapman and Hall, London.
Mayer, C., Davis, H., (1999), the Effect of the Performance Appraisal System on Trust for Management: A Field Quasi Experiment, Journal of Applied Psychology, Vol. 84, pp. 123-136.
References
-289-
Mayer, C., Davis, J., Schoorman, D., (1995), An Integrative Model of Organisational Trust, Academy of Management Journal, Vol. 20(3), pp. 709-734.
McAllister, D., (1995), Affect-and Cognition-Based Trust as Foundations for Interpersonal Cooperation in Organisations, Academy of Management Journal, Vol. 38 (1), pp.24-59.
McCarthy, E., (1964), Basic Marketing, Homewood, IL: Irwin.
McCullough, D., (1997), Quantitative vs. Qualitative, Marketing Research, [Online], Retrieved September 18, 1999 from http://www.macroinc.com/articles/
McKnight, D., Chervany N., (1996), The Meaning of Trust, Working Paper University of Minnesota, Minneapolis.
McKnight, D., Cummings, L., Chervany, N. L. (1998), Initial Trust Formation in New Organisational Relationships, The Academy of Management Review, Vol. 23(3), pp. 473-490.
McKnight, D., Chervany, N., (2000), The Meanings of Trust http://misrc.umn.edu/wpaper/WorkingPapers/9604.pdf McKnight, D., Chervany, N., (2001), The Meanings of Trust http://misrc.umn.edu/wpaper/WorkingPapers/9604.pdf
McKnight, D., Choudhury, V., Kacmar, C. (2002). Developing and Validating Trust Measures for E-commerce: an Integrative Typology, Information Systems Research, Vol. 13(3), pp. 334-359.
Messick, S., (1989), Research Validity, In R.L. Linn, Educational measurement, 3rd edition, pp. 13-103, New York: Macmillan.
Miles, B., Huberman, M., (1984), Qualitative Data Analysis: A Aourcebook of New Methods, London: Sage.
Milne, G., Boza, M., (1999), Trust and Concern in Consumers’ Perceptions of Marketing Information Management Practices, Journal of Interactive Marketing, Vol. 13(1), pp 87-97.
Misztal, B., (1996), Trust in Modern Societies, Polity Press: Cambridge.
Mohr, J., Spekman R., (1994), Characteristics of partnership success: partnership attributes, communication behavior and conflict resolution techniques, Strategic Management Journal, Vol.57 (January), pp. 135-152.
Möller, K., (2006), The Marketing Mix Revisited: Towards the 21st Century Marketing by E. Constantinides, Journal of Marketing Management, Vol. 22(3), pp. 439-450.
Möller, K., Halinen, A., (2000), Relationship Marketing Theory: Its Roots and Direction, Journal of Marketing Management, Vol. 16(1-3), pp. 29-54.
Moore, G., Sekhon, H., (2005), 1 Multi-Brand Loyalty Cards: A Good Idea, Journal of Marketing Management, Vol. 21(5/6), pp. 625-640.
References
-290-
Moorman, C, Zaltman, G., Deshpande, R., (1992), Relationships between Providers and Users of Market Research: The Dynamics of Trust within and between Organisations, Journal of Marketing Research, Vol. 29(3), pp. 314-329.
Moorman, C., Deshpande, R., Zaltman, G., (1993), Factors Affecting Trust in Marketing Research Relationships, Journal of Marketing, Vol. 57, pp. 81–101.
Morgan, R., Hunt S., (1994), The Commitment-Trust Theory of Relationship Marketing, Journal of Marketing, 58 (July), pp.20-38.
Mulaik, A., Millsap, E., (2000), Doing the Four-Step Right, Structural Equation Modeling, Vol. 7, pp. 36-73.
Nagelkerke, D., (1991): A Note on a General Definition of the Coefficient of Determination, Biometrika, Vol. 78, pp. 691–692.
Nau, D., (1995), Mixing Methodologies: Can Bimodal Research be a Viable Post-positivist Tool? The Qualitative Report, Vol. 2(3). From http://www.nova.edu/ssss/QR/QR2-3/nau.html
Netemeyer, R., Bearden, W., Sharma, S., (2003), Scaling Procedures: Issues and Applications, Sage Publications, Thousand Oaks, California.
Neuman, L., (1997), Social Research Methods: Qualitative and Quantitative Approaches, Boston: Allyn and Bacon.
Nielson, C., (1998), An Empirical Examination of the Role of ‘Closeness’ in Industrial Buyerseller Relationship, European Journal of Marketing, Vol.32(5/6), pp. 431-463.
Nooteboom, B., (1996), Trust, Opportunism and Governance: A Process and Control Model, Organisation Studies, 17 (6), pp. 985-1010.
Nooteboom, B., (2002), Trust: Forms, Foundations, Functions, Failures and Figures, Edward Elgar: Cheltenham.
Nooteboom, B., Berger, H., Noorderhaven, N., (1997), Effects of Trust and Governance on Relational Risk, Academy of Management Journal, 40 (2), pp. 308-338.
Nunnally, J., (1978). Psychometric Theory, New York: McGraw Hill.
O’Malley, L., Patterson, M., Evans, M., (1997), Retailer Use of Geo-demographic and Other Data Sources. International Journal of Retail and Distribution Management, Vol. 25(6), pp. 188–196.
O’Malley, L., Tynan, C., (1997), A Reappraisal of the Relationship Marketing Construct of Commitment and Trust, in Meenaghan, T. (Ed.), New and Evolving Paradigms: The Emerging Future of Marketing, Three American Marketing Association Special Conferences, pp. 486-503.
References
-291-
Oliver, R., (1999), Whence Customer Loyalty, Journal of Marketing, Vol. 63, Special Issue, pp. 33-44.
Pallant, J., (2007), SPSS Survival Manual, 3rd edition, Open University Press.
Palmer, A., (1996), Relationship Marketing: A Universal Paradigm or Management Fad, The Learning Organisation, Vol. 3(3), pp. 18-25.
Palmer, R., Lindgreen, A., Vanhamme. J., (2005), Relationship Marketing: Schools of Thought and Future Research Directions, Marketing Intelligence and Planning, Vol. 23(3), pp. 313-320.
Parasuraman, A., Zeithaml, A., Berry, L., (1988), SERVQUAL: a Multiple-Item Scale for Measuring Consumer Perceptions of Service Quality, Journal of Retailing, Vol. 64, spring, pp. 12-40.
Parvatiyar, A., Sheth, J., (1995), Relationship Marketing in Consumer Markets: Antecedents and Consequences, Journal of the Academy of Marketing Science, Vol. 23, pp. 255-71.
Parvatiyar, A., Sheth, J., (2000), The Domain and Conceptual Foundations of Relationship Marketing, in Sheth J.N., Parvatiyar A. (Eds.), Handbook of Relationship Marketing, Sage Publications Incorporated: Thousand Oaks, pp. 3-38.
Parvatiyar, A., Sheth, J., (2001), Customer Relationship Management: Emerging Practice, Process, and Discipline, Journal of Economic and Social Research, 3(2) 2001, pp. 1-34.
Parvatiyar, A., Sheth, J., (1997), Paradigm Shift in Inter-firm Marketing Relationships, Research in Marketing, Vol. 13, pp. 233-255.
Pavlou, P., (2001), Integrating Trust in Electronic Commerce with the Technology Acceptance Model: Model Development and Validation, in Strong, D., Straub, D., DeGross, J. (Eds),Proceedings of the Seventh Americas Conference on Information Systems, Association for Information Systems, Boston, MA, pp.818-20.
Payle, F., (1995), Humanism and Positivism in Nursing, Journal of Advanced Nursing, Vol. 22, pp. 979-984.
Pedhazur, J., Schmelkin, P., (1991), Measurement, Design, and Analysis: An Integrated Approach, Hillsdale: Lawrence Erlbaum Associates, Publishers.
Peterson, R., (1995), Relationship Marketing and the Consumer, Journal of the Academy of Marketing Science, Vol. 23, pp. 278-281.
Petrof, J., (March-April 1998), Relationship Marketing: the Emperor in Used Cloth, Business Horizons, pp. 79-82.
Piercy, N., (1995), Customer Satisfaction and the Internal Market: Marketing our Customers to our Employees, Journal of Marketing Practice: Applied Marketing Science, Vol. 1(1), pp. 22-44.
References
-292-
Piercy, N., Morgan, N., (1989), Internal Marketing Strategy: Leverage for Managing Marketing-Led Change, Irish Marketing Review, Vol. 4(3), pp. 11-28.
Piercy, N., Morgan, N., (1991), Internal Marketing the Missing Half of the Marketing Program, Long Range Planning, Vol. 24(2), pp. 82-93.
Pine, B., Peppers, D., Rogers, M., (1995), Do You Want to Keep Your Customers Forever? Harvard Business School Review, Vol. 2, pp. 103-114.
Platteau, J., (1994), Behind the Market Stage where Real Societies Exist, Part I: The Role of Public and Private Order Institutions, Journal of Development Studies, Vol. 30(3), pp. 533-560.
Polkinghorne, D., (1983), Methodology for the Human Sciences: Systems of Inquiry, Albany: New York Press.
Porter, W., Richard, S., Richard, M., Paul, B., (1974), Organisational Commitment, Job Satisfaction, and Turnover among Psychiatric Technicians, Journal of Applied Psychology, Vol. 59, pp. 603-609.
Rafiq, M., Ahmed, P.K., (1995), Using the 7Ps as a Generic Marketing Mix, Marketing Intelligence and Planning, Vol. 13(9), pp 4-15.
Reber, A., (1995), Implicit Learning and Tacit Knowledge: An Essay on the Cognitive Unconscious, Oxford University Press.
Reichheld, F., (1996), Learning from Customer Defections, Harvard Business Review, Vol. 74(2), pp. 56-69.
Reichheld, F., Sasser, W., (1990), Zero Defections: Quality Comes to Service, Harvard Business Review, pp.105-11.
Reichheld, F., Schefter, P., (2000), SE-Loyalty: Your Secret Weapon on the Web, Harvard Business Review, Vol. 78, pp. 10-13.
Reisinger, Y., Turner, L. (1999), A Cultural Analysis of Japanese Tourists: Challenges for Tourism Marketers, European Journal of Marketing; Vol. 33 (11/12), pp. 1203-1227.
Rempel, K., Holmes, G., Zanna, P., (1985), Trust in Close Relationships, Journal of Personality and Social Psychology, Vol. 49, pp. 95-112.
Ring, P., Van de Ven, A., (1992), Structuring Cooperative Relationships between Organisations, Strategic Management Journal, Vol. 13, pp. 483-498.
Ring, P., Van de Ven, A., (1994), Developmental Process of Cooperative Interorganisational Relationships, Academy of Management Review, Vol. 19, pp. 90-118.
Ring, S., (1996), Fragile and Resilient Trust and Their Roles in Economic Exchange, Business and Society, Vol. 35(2).
References
-293-
Robinson, S., (1996), Trust and Breach of the Psychological Contract, Administrative Science Quarterly, Vol. 41, pp.574-599.
Robson, C., (2002), Real World Research, 2nd edition, Blackwell Publishers: Oxford.
Rossiter, J, (2002), The C-OAR-SE Procedure for Scale Development in Marketing, International Journal of Research in Marketing, Vol. 19, pp. 305-335.
Rossiter, R., Pearch, L., (1996), Predictability and Trust, in Kramer, R., Tyler, T. (Eds.): Trust in organisations, frontiers of theory and research, Sage, Thousand Oaks, CA.
Rotter, J., (1971), Generalised Expectancies for Interpersonal Trust, American Psychologist, Vol. 26, pp.443-452.
Rotter, J., (1980), Interpersonal Trust, Trustworthiness, and Gullibility, American Psychologist, Vol. 35(1), pp. 1-7.
Rousseau, D., (1995), Psychological Contracts in Organisations, Sage, London.
Rousseau, M., Sitkin, S., Burt, R., Camerer, C., (1998), Not so Different After All: A Cross-Discipline View of Trust, Academy Movement Review, Vol. 23(3), pp. 393–404.
Rundle-Thiele, S., Bennett, R., (2001), A Brand for all Seasons? A Discussion of Brand Loyalty Approaches and their Application for Different Markets, Journal of Product and Brand Management, Vol. 10(1), pp. 25-37.
Rundle-Thiele, S., Mackay, M., (2000), Assessing The Performance of Brand Loyalty Measures, Journal of Services Market, Vol. 15, pp. 529-546.
Ruyter de, K., Wetzels, M., (1998), Determinants of Trust in Project Service Relationships, in Andersson P.(edited by), Market Relationships, proceedings of the 27th EMAC Conference, pp.99-110.
Sako, M., (1992), Prices, Quality and Trust: Inter-firm Relations in Britain and Japan, Cambridge: University Press.
Sarantakos, S., (1998), Social Research, 2nd edition, Macmillan Education Australia Ltd.
Sax, L., Bryant, A., Gillmartin, S., (2003) Assessing Response Rates and Non-response Bias in Web and Paper Surveys, Research in Higher Education, Vol. 44 (4), pp. 409-432.
Schumacker, E., Lomax G., (2004), A Beginners Guide to Structural Equation Modeling, Second Edition, Lawrence Erlbaum Associates, New Jersey.
Schurr, P., Ozanne, J., (1985), Influences on Exchange Processes: Buyers' Preconceptions of a Seller's Trustworthiness and Bargaining Toughness, Journal of Consumer Research, Vol. 11 (march), pp.939-953.
Selnes, F., (1998), Antecedents and Consequences of Trust and Satisfaction in Buyer-Seller Relationships, European Journal of Marketing, Vol.32, pp. 305-322.
References
-294-
Shani, D., Chalasani, S., (1992), Exploiting Niches Using Relationship Marketing, Journal of Consumer Marketing, Vol. 9(3), pp. 33-42.
Shapiro, P., (1987), The Social Control of Impersonal Trust, American Journal of Sociology, Vol. 93 (3), pp. 623-658.
Sheppard, L., Sherman, M., (1998), The Grammars of Trust: A Model and General Implications, Academy of Management Review, Vol. 23, pp. 422-437.
Sheppard, L., (1993), Evaluating Test Validity, Review of Research in Education, Vol. 19, pp. 405-450.
Sheth, N., Gardner, M., Garrett E., (1988), Marketing Theory: Evolution and Evaluation, John Wiley and Sons.
Shostack, G., (1977), Breaking Free From Product Marketing, Journal of Marketing, Vol. 41, April, pp. 73-80.
Siegrist, M., (2000), The Influence of Trust and Perceptions of Risks and Benefits on the Acceptance of Gene Technology, Risk analysis, Vol. 20, pp.195-203.
Sirdeshmukh, D., Singh, J., Sabol, B., (2002), Consumer Trust, Value, and Loyalty in Relational Exchanges, Journal of Marketing, Vol. 66 (1), pp. 15–37.
Sitkin, B., Roth, N, (1993), Explaining the Limited Effectiveness of Legalistic ‘Remedies’ For Trust-Distrust, Organisation Science, Vol. 4(3), p367-393.
Solomon, M., Surprenant, C., Czepiel, J., Gutman, E., (1985), A Role Theory Perspective on Dyadic Interactions: The Service Encounter, Journal of Marketing, Vol. 49(1), pp. 99-111.
Spekman, E., (1988), Strategic Supplier Selection: Understanding Long-Term Buyer Relationships, Business Horizons, July/ August, pp. 75-81.
Steenkamp, J.-B.E.M., Van Trijp H., (1991), The Use of LISREL in Validating Marketing Constructs, International Journal of Research in Marketing, Vol 12(8), pp.283-299.
Stone, L., Kotch, A., (1989), Individuals’ Attitudes Toward Organisational Drug Testing Policies and Practices, Journal of Applied Psychology, Vol. 74, pp. 518-521.
Stroh, M., (2000), Qualitative Interviewing, in Burton, D, Research Training for Social Scientists: A Handbook for Postgraduate Researchers, London: Sage Publications: London.
Steiger, J., Shapiro, A., Browne, M, (1985), On the Multivariate Asymptotic Distribution of
Sequential Chi-square Statistics, Psychometrika, Vol. 50, pp. 253-264.
References
-295-
Svensson, G., (2001), Extending Trust and Mutual Trust in Business Relationships Towards a Synchronised Trust Chain in Marketing Channels, Management Decisions, Vol. 39(6), pp. 431-440.
Swan, J., Trawick, I., Silva, D., (1985), How Industrial Salespeople Gain Customer Trust, Industrial Marketing Management, Vol. 14, pp.203-211.
Sweet, S., Martin, K., (2003), Data Analysis with SPSS: A First Course in Applied Statistics, Pearson Education.
Sztompka, P., (1999), Trust: a Sociological Theory, Cambridge University Press, Cambridge.
Tabachinck, B. Fidell, L., (1983), Using Multivariate Statistics, second edition, New York: Harper and Row.
Tabachnick, B., Fidell, L., (2007), Using Multivariate Statistics, 3rd edition, Prentice Hall.
Takala, T., Uusitalo, O., (1996), an Alternative View of Relationship Marketing: A Frame Work for Ethical Analysis, European Journal of Marketing, Vol. 30(2), pp. 45-60.
Tung, L., Tan, J., Chia, P.J.T., Yeo, L. (2001), An Empirical Investigation of Virtual Communities and Trust, Proceedings of ICIS 2001.
Turnbull, P., Cunningham, M., Ford, D., (1996), Interaction, Relationships and Networks in Business Markets: an Evolving Perspective, Journal of Business and Industrial Marketing, Vol. 131(3/4), pp. 44-62.
Tylor, R., Degoey, P., (1995), Collective Restraint in Social Dilemmas: Procedual Justice and Social Identifican Effect on Support for Authorities, Journal of Personality and social Psychology, Vol. 69(3), pp. 482-97.
Uncles, M., Dowling, G., Hammond, K., (2003), Customer Loyalty and Customer Loyalty Program, Journal of Consumer Marketing, Vol. 20(4), pp. 294-316.
Vargo, S., Lusch, R., (2004), Evolving to a New Dominant Logic for Marketing, Journal of Marketing, Vol. 68(2), 1-17.
Vaus, D., (1999), Surveys in Social Research, UCL Press: London.
Webb, J., (1996), Trust and Crisis, in Castaldo, S., (2007), Trust in Market Relationships, Edward Elgar, UK.
Webster, F., (1988), the Rediscovery of the Marketing Concept, Business Horizons, Vol. 31(3), pp.29-40.
Webster, F., (1992), The Changing Role in the Marketing in the Corporation, Journal of Marketing, Vol. 56(1), pp 1-18.
Webster, F., (1994), Executing the New Marketing Concept, Marketing Management, Vol. 3(1), pp. 9-18.
References
-296-
Wheaton, B., Muthen, B., Alwin, D., Summers, G. (1977), Assessing Reliability and Stability in Panel Models, Sociological Methodology, Vol. 8 (1), 84-136.
White, C., (1992), Identity and Control: A Structural Theory of Social Action, Princeton University Press.
Whitener, E., Brodt, S., Korsgaard, M., Werner, J., (1998), Managers as Initiators of Trust: An Exchange Relationship Framework for Understanding Managerial Trustworthy Behavior, Academy of Management Review, Vol. 23 (3), pp. 513-530.
Wicks, A., Berman, S., Jones, T., (1999), The Structure of Optimal Trust: Moral and Strategic Implications, Academy of Management Review, Vol. 24(1), pp. 99-116.
Williams, B., (1988), Formal Structures and Social Reality, in D. Gambetta (Ed) Trust: Making and Breaking Cooperative Relations, Oxford: Basil Blackwell, pp. 3-13.
Williams, J., Anderson, A., (1991), Job Satisfaction and Organisational Commitment as Predictors of Organisational Citizenship Behaviours and In-Role Behaviours, Journal of Management, Vol. 17(3), pp. 601-607.
Williamson, E., (1993), Calculative-ness, Trust and Economic Organization, Journal of Law and Economics, Vol. 30, pp. 131-145.
Williamson, O., (1985), The Economic Institutions of Capitalism, London: Free Press.
Yamagishi, T., Yamagishi, M., (1994), Trust and Commitment in the United States and Japan, Motivation and Emotion, Vol. 18, pp. 129-165.
Young, L., (2006), Trust: Looking Forward and Back, Journal of Business and Industrial Marketing, Vol. 21 (7), pp. 439–45.
Yuan, K., (2005), Fit Indices Versus Test Statistics, Multivariate Behavioral Research, Vol. 40 (1), pp. 115-48.
Zaheer, A., McEvily, B., Perrone, V., (1998), Does Trust Matter? Exploring the Effects of Inter-organisational and Interpersonal Trust on Performance, Organisation Science, Vol. 9, pp. 141-159.
Zaheer, A., Venkatraman, N., (1995), Relational Governance as an Interorganisational Strategy: An Empirical Test of the Role of Trust in Economic Exchange, Strcitegic, Ylanagcmenl Journal, Vol. 16, pp. 373-392.
Zaltman, G., Moorman, C., (1988), The management and Use of Advertising Research, Journal of Advertising Research, Vol. 28(6), pp.11-18.
Zand, E., (1972), Trust and Managerial Problem Solving, Administrative Science Quarterly, Vol. 17, pp. 229-239.
Zeithaml, A., Parasuraman, A., Berry, L., (1985), A Conceptual Model of Service Quality and Its Implications for Future Research, Journal of Marketing, Vol. 49(3), pp. 41-50.
References
-297-
Zeithaml, V., Bitner, M., (1996), Services Marketing, McGraw-Hill: UK.
Zikmund, G., Babin, J., (2007), Exploring Marketing Research, Thomson South-Western, UK.
Zineldin, M., (1999), Exploring the Common Ground of Total Relationship Management (TRM) and Total Quality Management (TQM), Management Decision, Vol. 37(9), pp. 719-728.
Zucker, L., (1986), Production of Trust: Institutional Sources of Economic Structure, 1840-1920, In Staw, B., Cummings, L., (Eds), Research in Organisational Behaviour, Greenwich, CT: JAI Press, Vol. 8, pp. 53-111.
Zwick, W., Velicer, W., (1986), Comparison of Five Rules for Determining the Number of Components to Retain. Psychological Bulletin, Vol. 99, pp. 432 - 442.
Appendices
-i-
APPENDICES
Appendix 1: The Study Questionnaire
Customer Evaluation of Hotel services Providers
Dear respondent:
The following questions relate with your personal evaluation of the current hotel you are dealing with. We would like some basic information about the relationship with the hotel we would like you to give us an indication of your perceptions of that hotel. Please can you spear approximately 5 minutes to complete this questionnaire. All information you provide will be treated in confidence.
Section I General Information
Q1 Can you please grade you MAIN HOTEL?
Q2 which of the following services do you use from your current hotel?
Accommodation Restaurant
Conference room Bar/Coffee shop
Health Club Banqueting
Other
Q3 How long you been a customer with this hotel?
Less than 1 year 1 - 2 years
3 - 5 years 6 - 10 years
11 - 15 years 16 - 20 years
21 - 25 years Over 26 years
1
3
5
2
4
6
7
1
3
5
7
2
4
6
8
5 1 2 4 4
Appendices
-ii-
Q4 How often you deal with the hotel named in Q1?
once a year 1 - 2 times/ years
3 – 5 times/ years 6 - 10 times/ years
First time Once every 3 years
Section II Your Assessment of The Hotel
Q1. Each of the following statements relate to your personal rating to the hotel you have named in Q1.
Please read each of the following statements and then indicate the extent to which you believe the statement to be relevant to your hotel. Please circle the number that mostly strongly reflects your feeling in relation to your hotel.
A My main hotel …
Stro
ngly
D
isag
ree
Neu
tral
Stro
ngly
A
gree
Show fairness in transactions 1 2 3 4 5 Always achieves consistency in service delivery 1 2 3 4 5 Always keeps its word 1 2 3 4 5 The hotel employees treat me with respect 1 2 3 4 5
B My main hotel …
Stro
ngly
D
isag
ree
Neu
tral
Stro
ngly
A
gree
Is open to my needs 1 2 3 4 5 Acts in a caring manner 1 2 3 4 5 Keeps its customers’ best interest in mind during most transactions
1 2 3 4 5
Is receptive to my needs 1 2 3 4 5
C My main hotel …
Stro
ngly
D
isag
ree
Neu
tral
Stro
ngly
A
gree
Has adequate skills to deliver the right service 1 2 3 4 5 Is efficient 1 2 3 4 5 The hotel employees can competently handle most of my requests
1 2 3 4 5
The hotel employees can be relied upon to know what they are doing
1 2 3 4 5
1
3
5
6
4
2
Appendices
-iii-
D My main hotel …
Stro
ngly
D
isag
ree
Neu
tral
Stro
ngly
A
gree
Shares the same values as me 1 2 3 4 5 Is ethical while dealing with me 1 2 3 4 5 Is interested in more than just making profit out of customers 1 2 3 4 5 The hotel employees act as if they value the customer 1 2 3 4 5
E My main hotel …
Stro
ngly
D
isag
ree
Neu
tral
Stro
ngly
A
gree
Has stability in its operations 1 2 3 4 5 Is always reliable 1 2 3 4 5 Always matches customers’ expectations 1 2 3 4 5 Always satisfies customers’ needs 1 2 3 4 5
F My main hotel …
Stro
ngly
D
isag
ree
Neu
tral
Stro
ngly
A
gree
Communicates clearly 1 2 3 4 5 Respond immediately when contacted 1 2 3 4 5 Inform me immediately of any problems 1 2 3 4 5 Always communicates with me 1 2 3 4 5
G My main hotel …
Stro
ngly
D
isag
ree
N
eutr
al
Stro
ngly
A
gree
I will transact with this hotel again for future visits 1 2 3 4 5 I will try new services that are provided by this hotel 1 2 3 4 5 I will recommend other people to visit to this hotel 1 2 3 4 5 I will say positive things to other people about the services provided at this hotel
1 2 3 4 5
H My main hotel …
Stro
ngly
D
isag
ree
Neu
tral
Stro
ngly
A
gree
I will continue to visit this hotel even if the service charges are increased moderately
1 2 3 4 5
I have strong preference to this hotel 1 2 3 4 5 I will keep visiting this hotel regardless of everything being changed somewhat
1 2 3 4 5
I am likely to pay a little bit more for using the services of this hotel
1 2 3 4 5
I My main hotel …
Stro
ngly
D
isag
ree
Neu
tral
Stro
ngly
A
gree
Appendices
-iv-
The hotel employees are professional and dedicated to customers
1 2 3 4 5
The hotel respond caringly when I share my problems 1 2 3 4 5 My hotel is always honest with me 1 2 3 4 5 Overall I feel I can trust my hotel 1 2 3 4 5 Q2 Please evaluate the following statements based on your personal view
K I Believe that………
Stro
ngly
D
isag
ree
Neu
tral
Stro
ngly
A
gree
I trust the hotel more if the hotel applies higher security standards
1 2 3 4 5
The hotel decorations are very important to me 1 2 3 4 5 The hotel atmosphere reflects the level of the expected service
1 2 3 4 5
The price of the hotel indicates its service level 1 2 3 4 5 The hotel brand name is very important to me 1 2 3 4 5 I trust my favourite hotel brand 1 2 3 4 5
Section III Personal Information
Q1 Gender
Male Female
Q2 Age group
Under 20 20-25
26-35 36-45
46-55 56-65
66+
Q3 Nationality
Thank you for your time
1
2
1
2
5
3
7
4
6
Appendices
-v-
Appendix 2: Review of Services Marketing Literature
Author Argument Proposition Booms and Bitner (1981)
Recognising the special character of the services as products, they demonstrated the importance of Environmental factors (Physical Evidence) influencing the quality perception. They included the Participants (personnel and customers) and the Process of service delivery as the additional Marketing Mix factors.
The Services Marketing Mix includes next to the 4Ps three more P’s: Participants Physical Evidence Process
Cowell (1984)
Three aspects justifying the revision of the original Marketing mix framework: the original mix was developed for manufacturing companies empirical evidence suggesting that marketing practitioners in the service sector find the marketing mix not being inclusive enough for their needs
Adopts the framework proposed by Booms and Bitner (1981).
Ruston and Carson (1989)
The unique characteristics of the services - intangibility, inseparability, perishability and variability - make the control of the marketing process, using the generalised tools of marketing, inadequate
New instruments and concepts must be developed to explain and manage the services intangibility
Fryar (1991)
Segmentation and differentiation is the basis of successful positioning of services. Furthermore the personal relationship with the customer and the quality of the service are important elements of the services Marketing
The Marketing of services requires: Differentiation based on segmentation and positioning Customer contact Unique vision on quality
Heuvel (1993)
Interaction between the one delivering the service and the customer is very important and has direct effect on the service quality and quality perception. The Product element can be better demonstrated as having two components, the primary and secondary service elements as well as the process
The Services Marketing Mix: • Personnel • Product Place • Price • Promotion
Doyle (1994)
While recognising that the content of the 4Ps in the service sector is somehow different
from that of the tangibles he does accept the 4Ps as the elements of the services marketing
mix. He identifies special difficulties in Promotion and Place preferring to replace
them by the terms Communication and Distribution
Service Marketing Mix: • Product • Price • Communication • Distribution
Melewar and Saunders (2000)
The Corporate Visual Identity System (CVIS) is the basis of the corporate differentiation and the core of the company’s visual identity.
A new P must be added to the 4Ps of the Marketing Mix (and the 3Ps of the Services Mix) namely the Publications
English (2000)
The traditional Marketing has never been an effective tool for health services marketing
A new framework emerges, emphasising the 4 Rs: Relevance, Response, Relationships, and Results.
Grove et al. (2000)
Services Marketing can be compared to a theatrical production. How the service is performed is as important as what is
Four strategic theatrical elements constitute the Services Experience: Actors,
Appendices
-vi-
performed. Critical factor is therefore the customer experience. The traditional Marketing Mix does not adequately capture the special circumstances that are present when marketing a service product
Audience, Setting, and Performance These elements must be added to the extended services Marketing Mix model of Booms en Bitner The four keys of Modern
Beckwith (2001)
Marketing services in a changing world requires focusing on increasing the customer satisfaction and rejecting old product paradigms and marketing fallacies.
The four keys of Modern (services) Marketing: Price, Brand, Packaging, and Relationships
(Source: Lee Goi 2009: 11)
Appendix 3: An Illustration of the Marketing Concept (appendices 3-9 removed for copyrightreasons)
(Kotler et al 2003:13)
Appendices
-vii-
Appendix 4: Strengths and Weakness of the 4Ps and 7Ps
(Source: Lee Goi 2009: 15)
Appendices
-viii-
Appendix 5: Review of Consumer Marketing Theory Literature
(Source: Lee Goi 2009: 8)
Appendices
-ix-
Appendix 6: Review of Relationship Marketing Literature
(Source: Lee Goi 2009: 9)
Appendices
-x-
Appendix 7: Sources of Communication Messages in a Relationship
Appendix 8: Route to a Relationship Marketing Concept
Source: (Gummesson 1998:244)
Appendices
-xi-
Appendix 9: Categories of Loyalty
Appendices
-xii-
Appendix 10: Judges Identification
Judge position Country One Academic researcher UK Two General hotel
manager Jordan
Three Academic researcher UK Appendix 11: Statistical Terms and Definitions Term Definition Population The entire collection of items that is the focus of the empirical
investigation. The branch of Statistics called, provides us with ways to describe the characteristics of a given population by measuring each of its items and then summarizing the set of measures in various ways
Descriptive Statistics
A collection of statistical tests providing ways to describe the characteristics of a given population by measuring each of its items and then summarising the set of measures in various ways, for example frequency analysis.
The mean The average of the scores in the population. Numerically, it equals the sum of the scores divided by the number of scores.
The median A test indicates the point on the scale of measures where the population is cantered. half of the scores in a population will have values that are equal to or larger than the median and half will have values that are equal to or smaller than the median
The variance A test used to characterise the spreading among the measures in a given population. It is necessary to first calculate the mean of the scores, then measure the amount that each score deviates from the mean and then square that deviation (by multiplying it by itself). Numerically, the variance equals the average of the several squared deviations from the mean.
The standard deviation
A test used to characterise the spreading among the measures in a given population. Numerically, the standard deviation is the square root of the variance. Unlike the variance, which is a somewhat abstract measure of variability, the standard deviation can be readily conceptualised as a distance along the scale of measurement.
The Standard Error
The Standard Error, or Standard Error of the Mean, is an estimate of the standard deviation of the sampling distribution of means, based on the data from one or more random samples. Numerically, it is equal to the square root of the quantity obtained when s squared is divided by the size of the sample
The null hypothesis
Is a term that often used to indicate the statistical hypothesis tested, with a purpose of deciding if whether the obtained results
Appendices
-xiii-
provide a reason to reject the hypothesis. For example, in an experiment in which two groups of randomly selected subjects have received different treatments and have yielded different means, it is always necessary to ask if the difference between the obtained means is among the differences that would be expected to occur by chance whenever two groups are randomly selected. In this example, the hypothesis tested is that the two samples are from populations with the same mean. Another way to say this is to assert that the investigator tests the null hypothesis that the difference between the means of the populations from which the samples were drawn, is zero. If the difference between the means of the samples is among those that would occur rarely by chance when the null hypothesis is true, the null hypothesis is rejected and the investigator describes the results as statistically significant.
Regression Given a pair of related measures ( X and Y ) on each of a set of items, the term "regression" is used to characterize the manner in which one of the measures (for example the Y measures) change as the other measure ( in this case, the X measure) changes
Chi Square a calculated value of Chi Square compares the frequencies of various kinds (or categories) of items in a random sample to the frequencies that are expected if the population frequencies are as hypothesised by the investigator
Latent variable An unobserved variable that may account for variation in the data and/or for apparent relations between observed variables
Scale reiability a measurement is reliable if it reflects mostly true score, relative to the error
Appendices
-xiv-
Appendix 12: The Calculation of the Composite Reliability
Composite reliability CR = (sum of standardised loading) 2 / (sum of
standardised loading) 2 + sum of indicator measurement error)
Integrity (IN) =(0.59+0.819+0.579+0.565)*2/((0.59+0.819+0.579+0.565)*2+(1-0.127+1-0.098+1-0.102))
Benevolence (BN) =(0.563+0.78+0.637+0.704)*2/((0.563+0.78+0.637+0.704)*2+(1-0.131+1-0.111+1-0.131))
Competence (CP) =(0.64+0.781+0.593+0.719)*2/((0.64+0.781+0.593+0.719)*2+(1-0.099+1-0.086+1-0.089))
Value Alignment (VA) =(0.578+0.859+0.641+0.775)*2/((0.578+0.859+0.641+0.775)*2+(1-0.119+1-0.102+1-0.103))
Consistency (CO) =(0.634+0.573+0.772+0.717)*2/((0.634+0.573+0.772+0.717)*2+(1-0.084+1-0.099+1-0.092))
Communication (CM) =(0.654+0.758+0.618+0.749)*2/((0.654+0.758+0.618+0.749)*2+(1-0.089+1-0.076+1-0.091))
Behavioural Loyalty (BL)
=(0.677+0.839+0.617+0.823)*2/((0.677+0.839+0.617+0.823)*2+(1-0.078+1-0.068+1-0.075))
Attitudinal Loyalty (AL)
=(0.71+0.609+0.717+0.636)*2/((0.71+0.609+0.717+0.636)*2+(1-0.072+1-0.085+1-0.08))
Trustworthiness (TW) =(0.576+0.677+0.815+0.572)*2/((0.576+0.677+0.815+0.572)*2+(1-0.108+1-0.132+1-0.098))
Appendices
-xv-
Appendix 13: Construct Reliability Evaluation 1. Integrity Reliability Statistics Cronbach's Alpha N of Items .730 4 Item-Total Statistics
Scale Mean if Item Deleted
Scale Variance if Item Deleted
Corrected Item-Total Correlation
Cronbach's Alpha if Item Deleted
Show fairness in transactions 13.1909 2.556 .475 .697
Always achieves consistency in service delivery
13.1796 2.598 .489 .688
Always keeps its word 13.0870 2.352 .633 .603 The hotel employees treat me with respect 13.0851 2.582 .491 .687
2. Benevolence Reliability Statistics Cronbach's Alpha N of Items .765 4 Item-Total Statistics
Scale Mean if Item Deleted
Scale Variance if Item Deleted
Corrected Item-Total Correlation
Cronbach's Alpha if Item Deleted
Is receptive to my needs 12.8091 3.284 .593 .695 Acts in a caring manner 12.7051 3.333 .542 .721 Keeps its customers' best interest in mind during most transactions
12.8261 3.038 .638 .668
Is receptive to my needs 12.7505 3.475 .489 .749
Appendices
-xvi-
3. Competence Reliability Statistics Cronbach's Alpha N of Items .776 4 Item-Total Statistics
Scale Mean if Item Deleted
Scale Variance if Item Deleted
Corrected Item-Total Correlation
Cronbach's Alpha if Item Deleted
Has adequate skills to deliver the right service 12.7543 3.261 .616 .702
Is efficient 12.6919 3.448 .508 .757 The hotel employees can competently handle most of my requests
12.7845 3.060 .654 .680
The hotel employees can be relied upon to know what they are doing
12.6957 3.379 .540 .741
Appendices
-xvii-
4. Value Alignment Reliability Statistics Cronbach's Alpha N of Items .801 4 Item-Total Statistics
Scale Mean if Item Deleted
Scale Variance if Item Deleted
Corrected Item-Total Correlation
Cronbach's Alpha if Item Deleted
Shares the same values as me 12.4991 3.868 .673 .723
Is ethical while dealing with me 12.3384 4.001 .548 .783
Is interested in more than just making profit out of customers
12.5104 3.542 .729 .691
The hotel employees act as if they value the customer
12.4216 4.169 .515 .797
5. Consistency Reliability Statistics Cronbach's Alpha N of Items .768 4 Item-Total Statistics
Scale Mean if Item Deleted
Scale Variance if Item Deleted
Corrected Item-Total Correlation
Cronbach's Alpha if Item Deleted
Has stability in its operations 12.4405 3.868 .546 .725
Is always reliable 12.3913 4.098 .498 .749 Always matches customers' expectations 12.4272 3.628 .638 .675
Always satisfies customers' needs 12.3403 3.782 .596 .699
Appendices
-xviii-
6. Communication Reliability Statistics Cronbach's Alpha N of Items .789 4 Item-Total Statistics
Scale Mean if Item Deleted
Scale Variance if Item Deleted
Corrected Item-Total Correlation
Cronbach's Alpha if Item Deleted
Communicates clearly 12.5879 3.262 .626 .722 Respond immediately when contacted 12.5236 3.686 .551 .760
Inform me immediately of any problems 12.6843 3.296 .635 .718
Always communicates with me 12.6427 3.423 .580 .746
7. Behavioural Loyalty Reliability Statistics Cronbach's Alpha N of Items .827 4 Item-Total Statistics
Scale Mean if Item Deleted
Scale Variance if Item Deleted
Corrected Item-Total Correlation
Cronbach's Alpha if Item Deleted
I will transact with this hotel again for future visits
12.7032 3.497 .714 .753
I will try new services that are provided by this hotel
12.6163 3.923 .560 .821
I will recommend other people to visit to this hotel
12.7618 3.386 .730 .745
I will say positive things to other people about the services provided at this hotel
12.6919 3.623 .612 .801
Appendices
-xix-
8. Attitudinal Loyalty Reliability Statistics Cronbach's Alpha N of Items .763 4 Item-Total Statistics
Scale Mean if Item Deleted
Scale Variance if Item Deleted
Corrected Item-Total Correlation
Cronbach's Alpha if Item Deleted
I will continue to visit this hotel even if the service charges are increased moderately
12.8450 2.828 .594 .689
I have strong preference to this hotel 12.8526 3.088 .519 .729
I will keep visiting this hotel regardless of everything being changed somewhat
12.9263 2.728 .596 .687
I am likely to pay a little bit more for using the services of this hotel
12.9319 2.851 .540 .719
Appendices
-xx-
9. Trustworthiness Reliability Statistics Cronbach's Alpha N of Items .756 4 Item-Total Statistics
Scale Mean if Item Deleted
Scale Variance if Item Deleted
Corrected Item-Total Correlation
Cronbach's Alpha if Item Deleted
The hotel employees are professional and dedicated to customers
12.6881 3.514 .521 .715
The hotel respond caringly when I share my problems
12.6181 3.437 .550 .700
My hotel is always honest with me 12.6522 3.102 .646 .643
Overall I feel I can trust my hotel 12.5085 3.648 .494 .729